Citation: 66 Fla. L. Rev. 1803 2014
Content downloaded/printed from HeinOnline
Fri Jul 21 11:28:22 2017
-- Your use of this HeinOnline PDF indicates your acceptance
of HeinOnline's Terms and Conditions of the license
agreement available at http://heinonline.org/HOL/License
-- The search text of this PDF is generated from
uncorrected OCR text.
-- To obtain permission to use this article beyond the scope
of your HeinOnline license, please use:
Copyright Information
Florida Law Review
Founded 1948
VOLUME 66 SEPTEMBER 2014 NUMBER 5
"SOPHISTICATED ROBOTS": BALANCING LIABILITY,
REGULATION, AND INNOVATION
F. PatrickHubbard
Abstract
Our lives are being transformed by large, mobile, "sophisticated
robots" with increasingly higher levels of autonomy, intelligence, and
interconnectivity among themselves. For example, driverless
automobiles are likely to become commercially available within a
decade. Many people who suffer physical injuries from these robots will
seek legal redress for their injury, and regulatory schemes are likely to
impose requirements on the field to reduce the number and severity of
injuries.
This Article addresses the issue of whether the current liability and
regulatory systems provide a fair, efficient method for balancing the
concern for physical safety against the need to incentivize the
innovation that is necessary to develop these robots. This Article
provides context for analysis by reviewing innovation and robots'
increasing size, mobility, autonomy, intelligence, and interconnections
in terms of safety-particularly in terms of physical interaction with
humans-and by summarizing the current legal framework for
addressing personal injuries in terms of doctrine, application, and
underlying policies. This Article argues that the legal system's method
of addressing physical injury from robotic machines that interact closely
with humans provides an appropriate balance of innovation and liability
for personal injury. It critiques claims that the system is flawed and
needs fundamental change and concludes that the legal system will
continue to fairly and efficiently foster the innovation of reasonably safe
sophisticated robots.
* Ronald L. Motley Distinguished Professor of Tort Law, University of South Carolina.
A preliminary version ofthis Article was presented as a paper at the WeRobot 2012 Conference
at the University of Miami School of Law. The author appreciates helpful comments on a draft
of this Article by Donna A. Dulo, Robert Felix, John Montgomery, and David Owen and the
research assistance of Candle Wester. Professor Jason O'Kane of the University of South
Carolina College of Engineering and Computing provided useful guidance concerning robotics
in the early stages of research for this Article.
FLORIDA LAWREVIEW
INTRODUCTION .............................................1806
1. "SOPHISTICATED ROBOTS" .................................................. 1806
A. Size, Mobility, Autonomy, Intelligence,and
Interconnection............................................................. 1806
B. DesignandSafety ......................................................... 1808
C. Innovation..................................................................... 18 10
II. APPROACHES FOR ADDRESSING SAFETY
AND PRODUCTS .................................................................... 1811
A. Liability: CorrectiveJustice/Market
Approach ........................................1812
1. Contract-Uniform Commercial Code
(UCC) .......................................1817
2. Tort ........................................................................ 18 19
a. Basic Concepts ......... ...................................... 1819
b. Products Liability: Sales,
Leases, and Other Distributions ..................... 1820
i. Doctrine.................................................... 1820
(a) The Three Types
of Defects.................. 1820
(b) Foreseeability, Misuse,
and Obvious Risks ............................ 1823
(c) Strict Liability................................... 1824
ii. Practice ..................................................... 1826
c. Reasonable Care for Control,
Use, and Maintenance .................................... 1829
i. Examples................................................. 1830
(a) Employment..................................... 1830
(b) Premises Liability ............................ 1831
ii. Exception for Abnormally
Dangerous Activities ............................... 1833
d. The Role of Standards ....... ......... 1834
e. Litigation Involving Robotic
Machines ........................................................ 1838
i. Early "Automatic"
Machines ................................................. 1838
ii. Recent Cases ........................................... 1839
(a) Industrial Robots.............................. 1839
(b) Automobiles..................................... 1840
(c) Da Vinci Surgical
"Robots" ......................... 1841
B. Standards:Legislative/Regulatory
Approach...................................................................... 1843
[Vol. 661804
"SOPHISTICATED ROBOTS"
1. Motor.Vehicles ...................................................... 1843
a. Standards ........................................................ 1843
i. Federal ...................................................... 1843
ii. State .......................................................... 1845
b. Other Impacts ................................................. 1846
2. Unmanned Aircraft System s (UAS) ...................... 1847
3. General-Purpose Robots ........................................ 1849
III. APPLICATION OF THE LIABILITY SYSTEM
TO SOPHISTICATED ROBOTS ................................................. 1850
A. SophisticatedAutomobiles with Human
Drivers.......................................................................... 1851
1. Sales, Leases, and Other
Distributions .......................................................... 1851
a. Complexity ..................................................... 1851
b. Software ......................................................... 1853
c. Design, Warnings, and
Instructions ..................................................... 1855
d. Trials .............................................................. 1857
e. Doctrinal Changes .......................................... 1857
2. Control, Use, and Maintenance ............................. 1858
3. Distributional Impacts ........................................... 1858
B. Fully Autonomous Sophisticated
Robots-EliminatingtheHuman in
the Loop ........................................................................ 1860
1. Traditional Doctrine-Changes
in Application ........................................................ 1861
2. Possible Doctrinal Expansions .............................. 1862
IV. PROPOSALS FOR FUNDAMENTAL CHANGE ............................ 1865
A. CompensatingVictims .................................................. 1866
1. No-Fault First Party Insurance
Schemes................................................................. 1866
2. No-Fault Third Party Liability
Schemes................................................................. 1867
B. FosteringInnovation: "Subsidies"
by Reducing Liability.................................................... 1869
1. Immunity .......................................1870
2. Preemption ............................................................ 1871
CONCLUSION ....................................................................................... 1872
1805
FLORIDA LAWREVIEW
INTRODUCTION
Within the next decade or so, our daily lives will be transformed by
"sophisticated robots," which will have much higher levels of
autonomy, intelligence, and interconnectivity than current robots. Many
of these machines will be sufficiently large and mobile enough to cause
physical injury and death, and people who suffer these physical injuries
may seek legal redress for their injury. In addition, regulatory schemes
are likely to impose requirements to reduce the number and severity of
these injuries. This Article addresses the issue of whether the current
approach of the legal system provides a fair, efficient method for
balancing the concern for physical safety against the need to support
and incentivize the innovation that is necessary to develop these
transformative robots.
Part I of this Article provides context for this argument by briefly
discussing innovation and robots' increasing mobility, autonomy,
intelligence, and interconnections in terms of safety-particularly in
terms of physical interaction with humans. Part II summarizes the
current legal framework for addressing personal injuries to illustrate the
framework's ability to achieve an efficient balance of innovation and
liability for personal injury. Part III addresses ways the legal system
may respond to address physical injury from large, mobile, and
sophisticated robots that interact closely with humans. Part IV critiques
proposals for fundamental change, which are based on claims that the
current liability system will not adequately compensate victims or that
potential liability costs under the current system inhibit robotics
innovation. Part V concludes that the legal system fairly allocates the
costs of injuries from innovation in robots and has not unduly hindered
innovation in robotics.
I. "SOPHISTICATED ROBOTS"
A. Size, Mobility,Autonomy, Intelligence,andInterconnection
There is no generally accepted definition of "robot" or "robotic."1
1. The term "robots" originated in a science fiction play, R.UR. (Rossum's Universal
Robots), by Karel t(apek, Karel tapek, R.UR. (Rossum's Universal Robots), reprintedin
TOWARD THE RADICAL CENTER: A KAREL tAPEK READER 34 (Peter Kussi ed., 1990), which
premiered in 192 1. Chronology,in TOWARD THE RADICAL CENTER, supra.In its original Czech
version, the play used the term "robota," which means "heavy labor." R.UR. (Rossum's
Universal Robots) The Makropulous Secret Inventions, in TOWARD THE RADICAL CENTER,
supra,at 32, 33. The movie Metropolisalso featured a robot, which was referred to in the movie
as a "machine man," though it was actually a female humanoid robot. METROPOLIS (Universum
Film AG 1927). The artificial entities in (Capek's play are more aptly viewed as organic
artifacts. See (apek, R.UR. (Rossum's UniversalRobots), supra, at 38-42 (depicting artificial
entities grown from organic living matter, engineered, and redesigned for mass production). The
term "robot" has come to refer to machines. See THE VISUAL ENCYCLOPEDIA OF SCIENCE FICTION
[Vol. 66
"SOPHISTICATED ROBOTS"
This Article views robots or robotic machines broadly as tools or
machines in terms of five characteristics: (1) size; (2) mobility; (3)
connectivity, in the sense that the machine can receive and transmit
information; (4) "autonomy" to respond to outside input by
independently engaging in physical motions; and (5) "intelligence,"
which refers to the rate at which the machine can receive, evaluate, use,
and transmit information, and the extent, if any, to which it can learn
from experience and use this learning in determining future responses.
Common examples of such robotic machines include the Roomba
"vacuuming robots" produced by the iRobot Corporation. 2
This characterization of robots is sufficiently broad to include a wide
range of autonomous machines, including common things like a cruisecontrol
system "driving" an automobile, an autopilot system in an
airplane, and perhaps even a "programmable," electronic thermostat
providing control for a heating and air conditioning system as it
responds to temperature changes. It would also include speculative
artificial intelligence systems like "HAL" (Heuristically programmed
ALgorithmic computer) in the science fiction movie 2001: A Space
Odyssey.3 Though HAL will remain a matter for science fiction in the
near future, it appears likely that, within the next decade or so, people
will live with a new class of "sophisticated robots" that differ radically
from current robots, not only in terms of their autonomy, intelligence,
and interconnectivity, but also in terms of increased size and mobility.4
172 (Brian Ash ed., 1977) (explaining that robots may be defined as entities, often made of
metal, whose minds are mechanical devices). Robots can take many forms, but humanoid robots
are the most popular form in science fiction. See id. at 175-80 (discussing the abundance of
science fiction stories featuring humanoid robots). Such robots are sometimes termed
"androids," but some writers restrict the term android to humanoid robots with synthetic
biological or chemical components that are grown rather than a humanoid mechanical entity that
is manufactured. See id. at 172, 180. As indicated above, the robots in Capek's play were such
synthesized androids. For further discussion of the development of the term "robot," see Jana
Hordkovd & Josef Kelemen, The Robot Story: Why Robots Were Born and How They Grew Up,
in THE MECHANICAL MIND IN HISTORY 283-306 (Philip Husbands et al. eds., 2008). For
historical examinations of robot myth and technology, see LISA NoCKs, THE ROBOT: THE LIFE
STORY OF A TECHNOLOGY xvii-xx, 3-19 (2007); see also SIDNEY PERKOWITZ, DIGITAL PEOPLE:
FROM BIONIC HUMANS TO ANDROIDS 17-84 (2004); Lev Grossman, lron Men and Bionic
Women, in RISE OF THE ROBOTS 18, 18-23 (Richard Stengel et al. eds., 2013).
2. Our History, IROBOT, http://www.irobot.com/us/Company/About/OurHistory.aspx
(last visited Aug. 15, 2014). The iRobot Corporation's website indicates that more than ten
million of its home robots had been sold as of 2013. Id. For further discussions of robots in the
home, see, for example, GREGORY BENFORD & ELISABETH MALARTRE, BEYOND HUMAN: LIVING
WITH ROBOTS AND CYBORGS 136 (2007) (discussing robotic household helpers) and DAVID
LEVY, LOVE + SEX WITH ROBOTS: THE EVOLUTION OF HUMAN-ROBOT RELATIONSHIPS 97-104
(2007) (discussing virtual and robotic "pets"); Michael Q. Bullerdick, Home Is Where the
HardwareIs, in RISE OF THE ROBOTS, supranote 1, at 48-51.
3. 2001: A SPACE ODYSSEY (MGM et al. 1968).
4. See, e.g., YOSEPH BAR-COHEN & DAVID HANSON, THE COMING ROBOT REVOLUTION
(2009) (discussing the future emergence of humanlike robots that will have more sophisticated
B. Design andSafety
Robots that might cause serious bodily injury are currently designed
in a way that is analogous to strict versions of the three "laws" of
robotics developed by Isaac Asimov:
1. A robot may not injure a human being, or, through
inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings
except where such orders would conflict with the First
Law.
3. A robot must protect its own existence as long as such
protection does not conflict with the First or Second
Law.5
Asimov used these laws as plot devices involving robots, rather than
as a detailed protocol for robot design. Though they are not actually
used in designing robots,6 crude forms of Asimov's laws are reflected in
current practice. More specifically, the first law is reflected in design
features that give robots little, if any, mobility and that isolate
dangerous robots from humans or cause a robot to stop activity if a
human comes within a danger zone.7 In terms of design, a combination
of the first and second laws parallel the wide use of the human-in-theloop
approach in such things as cruise control in cars, autopilot systems
in airplanes, 8 regulations concerning driverless vehicles on highways,9
features than the robots of today); Lev Grossman, DroneHome, in RISE OF THE ROBOTS, supra
note 1, at 24-43 (discussing increasing use ofunmanned aircraft systems).
5. ISAAC ASIMOv, ROBOT VISIONS 8 (1990); see also IsAAc ASIMOv, THE NAKED SuN 31-
33 (1957).
6. See, e.g., RODNEY A. BROOKS, FLESH AND MACHINES: How ROBOTS WILL CHANGE US
73 (2002) ("[W]e do not know how to build robots that are perceptive enough and smart enough
to obey these three laws."); WENDELL WALLACH & COLIN ALLEN, MORAL MACHINES: TEACHING
ROBOTS RIGHT FROM WRONG (2009) (discussing challenges of whether and how to design robots
able to make complex decisions about "right" behavior); F. Patrick Hubbard, "Do Androids
Dream?": Personhood and Intelligent Artifacts, 83 TEMP. L. REV. 405, 464-66 (2011)
(discussing Asimov's robot novels and short stories, noting the role of the three laws as plot
device and noting that these laws cannot be used to program robots); Robin R. Murphy & David
D. Woods, Beyond Asimov: The Three Laws of Responsible Robotics, 24 IEEE INTELLIGENT
SYSTEMS, July-Aug. 2009, at 14, available at http://www.inf.ufrgs.br/-prestes/Courses/
Robotics/beyond%20asimov.pdf.
7. See infra notes 153-56 and accompanying text (discussing cases that illustrate lack of
success ofsuits for injuries from industrial robots).
8. See, e.g., Brouse v. United States, 83 F. Supp. 373, 374 (N.D. Ohio 1949) (holding
that operator of plane "under robot control" was negligent in failing to "keep a proper and
constant lookout" for other planes).
1808 FLORIDA LA W REVIEW [Vol. 66
"SOPHISTICATEDROBOTS"
and missile-firing decisions in drone aircraft. 10 Finally, the third law is
reflected in features that alert users of the need for recharging a battery
or that enable a machine to find a power source and recharge its battery
on its own.lI
The safety achieved by these design features comes at a high price
because their severe limits on autonomy drastically reduce the
usefulness of robots. Following the first law, by isolating robots from
humans by barriers or automatic stop features, this limits the ability of
humans to engage in collaborative efforts with potentially dangerous
robotic machines of far greater size and capabilities than a Roomba, in a
wide variety of settings. 2 The human-in-the-loop approach reduces the
ability of robots to replace humans in many tasks by only providing
human assistance. As a result, current design approaches inhibit the
development and use of robots with the autonomy of physical
movement that will enable their integration into daily life to perform
physical tasks in the same ubiquitous way that computers have come to
handle information.
Although the present approach to robot design results in a high level
of human safety, the limitations necessary to achieve this level raise a
question: Is society paying too high a price in foregone benefits for this
level of protection? Safety is, of course, important; but it is just one
social value. All technology presents the challenge of balancing its costs
9. See infra notes 190-201 and accompanying text (discussing the Nevada regulatory
scheme, which requires the presence of a human being in a vehicle and equipment that enables
the human being to take immediate control).
10. See, e.g., P.W. SINGER, WIRED FOR WAR: THE ROBOTICS REVOLUTION AND CONFLICT
INTHE TWENTY-FIRST CENTURY passim (2009); Peter Finn, A Futurefor Drones: Automated
Killing, WASH. POST- (Sept. 19, 2011), http://www.washingtonpost.com/national/nationalsecurity/a-future-for-drones-automated-killing/2011/09/15/glQAVy9mgK~story.html
(discussing
research on removing humans from the decision to kill). For more on the tension between the humanin-the-loop
approach and autonomy in the military context, see SINGER, supra,at 123-34.
11. See, e.g., Is Kiva Systems a Good Fitfor Your DistributionCenter?: An Unbiased
Distribution ConsultantEvaluation, MWPVL, http://www.mwpvl.com/html/kiva-systems.html
(last visited Aug. 15, 2014) (discussing Kiva warehouse robots that "travel to designated charge
stations every couple of hours where they receive a 5-minute battery re-charge").
12. See, e.g., Safety of Human-Robot CollaborationSystems Project, NAT'L INST. OF
STANDARDS & TECH. (Nov. 10, 2011), http://web.archive.org/web/20120925043230/
http://www.nist.gov/eI/isd/ps/safhumrobcollsys.cfin (archived copy) [hereinafter NIST, HumanRobot
Collaboration] (discussing safety standards that allow robots and humans to work
together in the same space); Safety ofHuman-RobotSystems in Flexible FactoryEnvironments,
NAT'L INST. OF STANDARDS & TECH. (Jan. 10, 2013), http://www.nist.gov/el/isd/ps/
safehumrobsysflexfactenvir.cfm (discussing the development of methods for modeling and
evaluating the performance of safety systems that will allow the collaboration of robots and
humans in factory environments). NIST notes: "Safe human-robot collaboration is widely seen
as key to the future ofrobotics. When humans and robots can work together in the same space, a
whole class of tasks becomes amenable to automation, ranging from collaborative assembly to
parts and material handling and delivery." NIST, Human-RobotCollaboration,supra.
1809
against its benefits. For example, because of the enormous benefits of
automobiles, there is no debate about whether to use them despite their
high cost, not only in terms of injuries and deaths, but also in terms of
urban sprawl, pollution, and dependence on foreign oil supplies. People
do try to reduce these costs; but most of us are not seriously considering
abandoning automobiles. Instead, we seek an efficient balance of their
costs and benefits.
Similarly, because of the benefits of robots with greater size,
mobility, connectivity, autonomy, and intelligence society will likely
relax the current rigid limitations on robots where it is possible to do so
in a cost-effective manner. The push for such relaxation of limitations
has already begun, partly because of developments in autonomous
vehicles. 13 This push is likely to expand into other areas-for example,
robots with increased abilities to identify humans and predict their
behavior "are under development throughout the world... [and] will
revolutionize manufacturing by allowing humans and robots to operate
in close proximity while performing a variety of tasks."'14
Such
capabilities will make it possible to enjoy the benefits resulting from
reducing the scale of the danger zone that requires a robot to cease
activity if humans come within the zone and from reducing the need for
a human in the loop. Consequently, as with automobiles, society may
come to accept the benefits of the technology despite the increased risk
of injuries to humans.
C. Innovation
Innovation will be necessary for the development of large mobile
sophisticated robots that can safely interact closely with humans.
Despite widespread agreement that technological innovation is valuable
and that innovation includes "improved products," it is not always clear
what counts as innovation.15
For example, inventions and patents are
often related to innovation, but most patents are never used and thus do
not result in new or improved technology.' 6 Similarly, there is no
agreement on the relative roles of government and the private market in
13. See, e.g., Grossman, supra note 1, at 24-33; see infra notes 180-216 and
accompanying text; infra Section IV.B. The "robot industry" has an active trade organization,
Association for Unmanned Vehicle Systems International (AUVSI), which coordinates the push
to develop unmanned systems. For information on the organization, see AUVSI,
http://www.auvsi.org/home/aboutus (last visited Aug. 15, 2014).
14. NIST, Human-Robot Collaboration,supranote 12.
15. NAT'L RESEARCH COUNCIL, IMPROVING MEASURES OF SCIENCE, TECHNOLOGY AND
INNOVATION: INTERIM REPORT 20-26 (Litan et al. eds., 2012).
16. William Hubbard, CompetitivePatentLaw, 65 FLA. L. REv. 341, 354 n.92 (2013); see
NAT'L RESEARCH COUNCIL, supranote 15, at 20-23.
1810 FLORIDA LA W REVIEW [Vol. 66
"SOPHISTICATED ROBOTS"
promoting innovation.'
7
This Article addresses uncertainty about the nature of innovation in
two ways. First, this Article considers innovation in terms of situations
where humans face increased risk of physical injury from the use of
sophisticated robots. Second, it considers innovation in terms of the
impact of liability for harm caused by a robot on innovators' decisions
to invest in developing these robots. Unfortunately, the second approach
simply restates the problem of defining innovation as a question of who
counts as an innovator. For example, are venture capitalists innovating
if they invest in a company that is developing innovative robots? To
further limit the scope of analysis, this Article will restrict the concept
of innovator to manufacturers, designers, sellers, lessors, and other
distributors of robots and their physical components. Thus, this Article
will focus on the effect of regulation and liability on these actors. In
addressing this effect, this Article assumes that these actors will base
their decisions on a rational comparison of the possible economic
benefits of success in innovation with the possible costs, including costs
of liability for injuries and of satisfying regulatory requirements. 1pII.
APPROACHES FOR ADDRESSING SAFETY AND PRODUCTS
In any area of technological innovation, the legal system faces the
challenges of: (1) reducing the number and severity of personal injuries;
(2) allocating the costs of victims' injuries;19 and (3) incentivizing
innovation. For more than two centuries, the United States has
addressed these concerns with a complex system of federal, state, and
local governmental entities, each of which: (1) is relatively autonomous,
and (2) uses both judicial and legislative/regulatory mechanisms to
address injuries. The system has changed enormously in response to
problems resulting from developments like mass production of goods,
17. See, e.g., Jeff Madrick, Innovation:The Government Was CrucialAfter All, N.Y. REv.
OF BOOKS 50 (Apr. 24, 2014) (arguing that federal government plays a crucial role in technological
innovation), available at http://www.nybooks.com/articles/archives/2014/apr/24/innovationgovernment-was-crucial-after-all/;
Jim Manzi, The New AmericanSystem, 19 NAT'L AFFAiRs (Spring
2014) (arguing that although government has had and will continue to have an important role in
innovation, the role is different in the current world), available at
http://www.nationalaffairs.com/publications/detail/the-new-american-system.
18. In practice, this comparison can be difficult and, at times, stressful. See infra notes
237-50 and accompanying text. Moreover, rationality in the face of uncertainty may be partly a
matter of whether a person adopts a maxi-max (maximize the maximum received if potential
gains exceed losses) or a maxi-min (maximize the minimum that results if things go very
poorly) strategy.
19. The allocation of loss is complicated because, in addition to compensation based on
liability in a corrective justice scheme, victims' injury costs can also be addressed by a wide
range of both private and social insurance schemes. See, e.g., F. Patrick Hubbard, The Nature
andImpactofthe "Tort Reform" Movement, 35 HOFSTRA L. REv. 437,441-42 (2006).
1811
FLORIDA LAW REVIEW
mechanized transportation systems, and electronic communications
systems.
The current system can be categorized in terms of two distinct
approaches to safety. The first uses a judicially imposed liability system
of corrective justice that requires wrongdoers to compensate victims for
injuries caused by a wrong. The goal is to create market incentives that
internalize the costs of wrongdoing to the wrongdoer. In this way,
individual private actors decide the most efficient way to address
potential liability for wrongful injury. The second approach involves
collective determinations of the best way to address safety and imposes
this determination with regulatory commands backed by the threat of
sanctions for violations. The two approaches often work in tandem. For
example, the rational driver of an automobile deciding how fast to drive
would consider the possibility of being held liable for injuries caused by
wrongfully excessive speed as well as the risk of receiving a speeding
ticket.2 °
A. Liability: CorrectiveJustice/MarketApproach
The legal system relies on the judicially administered system of
contract and tort liability-largely a matter of state law-to address
personal injury caused by robots. These injuries will be borne primarily
by three types of victims: (1) purchasers(owners) or lessees, (2) users
(other than purchasers or lessees), and (3) otherparties.The important
differences among these types of victims are: (1) only purchasers and
lessees can bargain with innovators for contract rights at the time of
sale, lease, or other distribution; and (2) purchasers, lessees, and users
will enjoy benefits offsetting the risks from the vehicles, while other
parties may incur only risks.
Both contract and tort law impose liability-based primarily on
"fault" or "wrongdoing"--for injuries caused by the use of machines.
This liability provides an incentive to avoid "wrongdoing" by making
safer products and using products more safely. As the development and
use of robots proceeds, products liability law, including both contract
and tort doctrines, will likely govern the risk of injuries from the sale,
lease, or other distribution of robots, and the general negligence
principles of tort law will primarily govern injuries from the use of such
products.
Contract and tort are "corrective justice" schemes for allocating the
risk of loss based on the following principle: Where Plaintiff (P) can
20. See infra note 25 for discussion of this point in terms ofefficiency.
21. Determining whether other parties receive benefits as well as risks can be complicated
in practice. For example, a person who does not own or drive a car benefits indirectly from the
use ofmotor vehicles.
1812 [Vol. 66
"SOPHISTICATED ROBOTS"
show that Defendant's (D's) wrongdoing (defined as a breach of a legal
duty to P) caused injury to P, D must correct that wrong by placing Pusually
through compensation-in the position P would have been in
but for the wrong by D.22
Thus, in both contract and tort, the plaintiff
bears the burden of proving a right to compensation from the defendant.
The primary difference between contract and tort is that legal duties are
generally imposed by agreement of the parties in contract and by
operation ofthe law in tort.23
As indicated more fully below, efficiency in terms of requiring only
cost-effective expenditures on safety plays a central role in defining the
duties to potential victims owed by product sellers, lessors, and users.24
As a result, corrective justice and the efficient reduction of accidental
injury costs are not necessarily in conflict with each other. Instead,
because efficiency generally defines duties, the corrective justice
systems of contract and tort tend to promote efficiency.25
Contract law and tort law are largely matters of state law. This
Article adopts two approaches to address the complexity resulting from
variations among the states concerning the details of legal doctrine.
First, this Article considers contract law primarily in terms of Article 2
of the Uniform Commercial Code (UCC), which virtually all states have
adopted (with occasional minor changes). Article 2 of the UCC governs
sales and other forms of distribution in terms of "transactions in
goods, 26 and Article 2A governs leases.2 7
Both articles apply torobots.28 Second, the discussions of tort law rely on sources such as
22. Traditionally, a claim or "cause of action" in tort or contract is viewed in terms of the
following "elements": (1) a duty owed to plaintiff by defendant; (2) a breach of that duty
("wrongdoing"); (3) injury caused by that breach; and (4) legally cognizable damages (or other
remedy) for the injury caused by the breach. See, e.g., DAN B. DOBBS, THE LAW OF TORTS § 114,
at 269 (2001). For general discussion of "corrective justice" and purposes of tort law, see, for
example, Hubbard, supranote 19, at 446-48.
23. Because tort and contract law are judicially administered, the impact of the
substantive rules in both legal areas is partly determined by damages rules, evidentiary rules
concerning proof, and procedural rules concerning trial. Discussion of these more general rules
is beyond the scope of this Article.
24. See infra notes 43, 54-56, 59-61, 250 and accompanying text.
25. Even though tort law incentivizes efficiency by only imposing liability for
inefficiency in making decisions about safety, it does not perform this function in the same way
as a regulatory system. For example, if a person acts in a negligently dangerous manner by
driving twenty miles per hour over the speed limit and no one is harmed as a result, there is no
wrong to correct, and thus, no liability in tort. In contrast, under regulatory schemes addressing
driving, a fine for that speeding could be imposed regardless of outcome and the amount of that
fine would generally be set regardless of whether harm occurred.
26. U.C.C. § 2-102 (2012); § 2-105 (defining "goods"); § 2-106(1) (defining "sale").
27. U.C.C. § 2A-102.
28. CompareU.C.C. § 2-102 ("[T]his Article applies to transactions in goods. ), with
§ 2-105 (defining "goods" as "all things (including specially manufactured goods) which are
movable at the time ofidentification").
widely used treatises and the Restatements of the Law of Torts,29
particularly the Restatement (Third) of Torts: Products Liability
(referred to hereinafter as Restatement of Torts. ProductsLiability)."
These simplifying techniques are supplemented by more detailed
discussion of specific examples of litigation concerning robots. This
approach does not restrict the utility ofthis analysis of the impact of the
liability system on innovation for three reasons.
First, persons considering whether to sell, buy, or develop robotic
products will evaluate risk ex ante-i.e., before any injuries have
occurred. From this perspective, they usually will not know where and
how the accident may occur. As a result, they will not know which
state's law will apply, and thus, will be concerned with the risk of
liability for injury in any state where an injury might occur. The U.C.C.
and the Restatements provide a useful basis for such a broad ex ante
consideration of risk.3'
29. The Restatements provide a neutral authoritative summary of basic doctrinal areas in
American law. They are prepared by the American Law Institute (ALl). The website for the ALl
describes its work as follows:
The American Law Institute is the leading independent organization in the
United States producing scholarly work to clarify, modernize, and otherwise
improve the law. The Institute (made up of 4000 lawyers, judges, and law
professors of the highest qualifications) drafts, discusses, revises, and publishes
Restatements of the Law, model statutes, and principles of law that are
enormously influential in the courts and legislatures, as well as in legal
scholarship and education.
ALl Overview, AMERICAN LAW INSTITUTE, http://www.ali.org/index.cfln?fuseaction
=about.overview (last visited Aug. 15, 2014).
30. The Product Liability Restatement was adopted by the ALl on May 20, 1997.
RESTATEMENT (THIRD) OF TORTS: PROD. LA-B. (1997).
31. Unfortunately, it is not possible to address other important areas of doctrine relevant
to allocating the risk of liability for injuries caused by sophisticated robots. In addition to the
omitted topics discussed in supra note 23, the following are not addressed: (1) defenses,
particularly those based on conduct by the victim; and (2) allocation of liability among
innovators (and to a lesser extent, among innovators, purchasers, and users). Though the
conduct ofusers and victims is addressed in the discussion of design and warnings at Subsection
II.A.2 below and the allocation involved in the second area is addressed to some extent in the
discussion of indemnity infra note 45, the other aspects of these doctrinal areas will not be
addressed herein. Covering these other areas is simply too complicated to be addressed in an
article of this length. Given, the ex ante nature of risk assessment, this necessary limitation on
coverage should not substantially affect analysis.
Similarly, broad rights to prevent harm to person or property by engaging in self-help or by
using another's robot are not addressed. For a discussion of such rights, see, for example,
Vincent v. Lake Erie Transp. Co., 124 N.W. 221, 222 (Minn. 1910) (requiring one who
exercises privilege of necessity to protect his property by use of another's property to
compensate the person whose property is used); Ploof v. Putnam, 71 A. 188, 189 (Vt. 1908)
(recognizing a privilege based on necessity to use another's property to protect one's property);
A. Michael Froomkin & Zak Colangelo, Self-Defense Against Robots 7-33 (Mar. 19, 2014)
1814 FLORIDA LA W REVIEW [Vol. 66
"SOPHISTICATED ROBOTS"
Second, the content of a legal rule or standard is always
indeterminate to some extent.32 This indeterminacy results partly from
the limits of verbal communication, which makes it impossible to devise
rules precise enough to decide cases outside of a "core of undisputed
meaning" containing a limited class of clear cases.33
In addition,
because of the dynamism and complexity of the world, lawmakers have
limited ability to determine present and future facts. As a result, they are
handicapped by a "relative indeterminacy of aim" concerning the
subject of the rule, and this indeterminacy presents difficulties in
interpreting and applying the rule.34 It may also result in a decision to
use a broad rule that, though often indeterminate in terms of specific
applications, is flexible enough to address diverse fact scenarios. 35
Analogous problems arise in programming autonomous robots to
perform complex tasks in unanticipated, changing environments.36
Finally, even relatively precise rules can be changed, and these changes
can apply even if they are enacted after the sale of the product.37
Third, doctrinal uncertainties are generally less important than the
wide variation in the types and amount of harms that might result when
a product is wrongfully designed or used. For example, where a failure
of an autonomous braking system causes an automobile collision, the
injuries caused by the collision could range from minor bruises to spinal
injury that renders the victim a quadriplegic. Though rare, serious
injuries like quadriplegia typically involve millions of dollars in
economic damages for a life care plan, as well as potentially large
(unpublished conference draft), available at http://robots.law.miami.edu/2014/wp-
content/uploads/2013/06/Froomkin-Colangelo-Self-Defence-Against-Robots-March-2014.pdf.
32. See H.L.A. HART, THE CONCEPT OF LAW 126, 144-45,272-73 (2d ed. 1994).
33. Id. at 12; see, e.g., id.at 123 ("core of certainty"); 128-29 ("paradigm, clear cases").
34. Id.at 128.
35. See infra note 57 and accompanying text.
36. See, e.g., BROOKS, supra note 6, at 19-21 (discussing "tortoise" robots and their
"remarkably unpredictable" responses to changing environments). See generally WALLACH &
ALLEN, supra note 6, at 83-124 (discussing relative strengths and weaknesses of top-down and
bottom-up approaches to designing robots able to handle change in complexity). Though
programs are not ambiguous, it is not possible to provide rules for all possible cases because of
problems involved in providing a complete model of the robot's environment. In addition,
difficulties can arise as a result of emergent behavior where a robot is given some autonomy.
See infra note 227 and accompanying text.
37. Statutory changes in law often apply to causes of actions [claims] "arising after" the
effective date of the statute. Typically, a cause of action arises after the event in which a product
caused the injury, regardless of whether the sale of the product occurred before the statutory
change. Judicial changes often follow a similar pattern. See, e.g., Marcum v. Bowden, 643
S.E.2d 85, 90-91) (S.C. 2007) (Toal, J., concurring in part, dissenting in part) (arguing that the
change in rule concerning tort liability should not be prospective for claimant bringing
successful challenge); Steinke v. South Carolina Dep't of Labor, Licensing, & Regulation, 520
S.E.2d 142, 156 (S.C. 1999); ROBERT E. KEETON, VENTURING TO Do JUSTICE: REFORMING
PRIVATE LAW 41, 51 (1969).
FLORIDA LA WREVIEW
awards for psychic harm.
Given all these uncertainties, the rational ex ante approach for
innovators is to find the best mix of basic product liability insurance and
self-insurance for their particular robotic product. Though the premium
may vary from year to year,38 the cost of this insurance package will
provide a relatively reliable figure as to the liability cost for innovation.
Unfortunately, however, because of the nature of products liability
insurance, things may be more complicated. Conceptually, liability
insurance is a simple matter: The insurer sets premium rates and invests
premiums in order to secure a sufficient income to cover administrative
expenses (including litigation costs), to pay claims, and to make an
adequate profit.39 Yet achieving this result is challenging because of
uncertainty concerning investment income and claims payouts. All
types of insurance face investment uncertainty and problems in
predicting the amount and timing of claims payouts. To some extent,
insurance companies can address the problem in predicting claims
where a large body of claims data exists, as in the areas of automobile
insurance 40 and general commercial liability.41 However, there may well
be no such data available to insurers where a seller seeks liability
insurance for an innovative sophisticated robot. As a result, products
liability insurers may be very concerned about the potential for high
claims. Therefore, insurance may be hard to get, very expensive, or
both.42
38. A full discussion of the details of insurance is beyond the scope of this Article. In
terms of details, cost variation in insurance can result from factors like increased (or decreased)
competition, reduced investment opportunities for the fund established to pay claims, changes in
administrative costs, and increased payouts for claims. Reductions in the amount of coverage in
terms of exclusions, types of claim coverage (occurrence or claims made), and dollar amount of
coverage are also important. See, e.g., KENNETH S. ABRAHAM, DISTRIBUTING RISK: INSURANCE,
LEGAL THEORY, AND PUBLIC POLICY 48 (1986).
39. See, e.g., id at 77.
40. See, e.g., id. at 46.
41. See, e.g., Mark A. Geistfeld, Legal Ambiguity, Liability Insurance,and Tort Reform,
60 DEPAUL L. REV. 539, 549-50 (2011) (explaining that while the insurance companies can get
large pools of data on general commercial liability, there are still factors that make accurately
predicting future liability costs difficult).
42. See id. at 552-64. For approaches to address this problem, see, for example, STEPHEN
S. Wu, RISK MANAGEMENT IN COMMERCIALIZING ROBOTS 6-15 (2013), available at
http://conferences.law.stanford.edu/werobot/wp-content/uploads/sites/29/2013/04/RiskManagement-in-Commercializing-Robotics.pdf
(presented at the Stanford Law School
Conference "We Robot: Getting Down to Business"). For useful discussion of insurance in
terms of Unmanned Aircraft Systems (UAS), see, for example, Breyer et al., Risk, Product
Liability Trends, Triggers, and Insurance in CommercialAerial Robots (2014), available at
http://robots.law.miami.edu/2014/program/.
1816 [Vol. 66
"SOPHISTICATED ROBOTS"
1. Contract-The Uniform Commercial Code (UCC)
A contract's allocation of risk is viewed as the result of rational
choices by autonomous agents, therefore the legal system generally
treats rights based on a contract as both fair and efficient.43 Two kinds
of contracts are important to the allocation of risk of personal injuries:
(1) contracts made Vursuant to UCC rules governing sales, leases, and
other distributions; 4 and (2) contracts of indemnification, which
explicitly allocate a risk of liability for loss by an agreement to shift the
liability from one party (the indemnitee) to the other party (the
indemnitor).45
The UCC has a number of default rules that structure the contracting
process. One of the most important default rules in terms of productcaused
injuries to persons is: Products sold or leased by "a merchant
[(seller)] with respect to goods of that kind" must be "merchantable,"
which means, among other things, "fit for the ordinary purpose for
43. The underlying goals of contract law are the subject of dispute, primarily in terms of
whether the goal is better expressed in terms of the promotion of efficiency or the protection of
promise-based or expectation-based rights. See STEPHEN A. SMITH, CONTRACT THEORY 3 (2004).
Compare RICHARD A. POSNER, ECONOMIC ANALYSIS OF LAW § 4.9 (6th ed. 2003) (arguing for
efficiency), with CHARLES FRIED, CONTRACT AS PROMISE: A THEORY OF CONTRACTUAL
OBLIGATION 8 (1981) (discussing promise-based rights), and HENRY MATHER, CONTRACT LAW
AND MORALITY 1, 3-6 (1999) (asserting that facilitation of reliance and beneficial coordination
is the goal). Under both approaches, however, the result is the same-i.e., the wishes of the
parties as expressed in the contract are generally enforced. On rare occasions, all or part of a
contract will be viewed as so unfair as to be unenforceable. See, e.g., U.C.C. § 2-302 (2012)
(providing that a judge may refuse to enforce contract (or clause in a contract) on grounds that
the contract or clause is "unconscionable"). The basic concern is "the prevention of "oppression
and unfair competition." Id. at cmt. 1.
44. See supra notes 26-27 and accompanying text. Where a contract has an express
warranty by the seller, the seller is liable regardless of fault for breach of that warranty. See
DAVID G. OWEN, PRODUCTS LIABILITY LAW 32 (2d ed. 2008).
45. A person who has been held liable for injury in tort or contract, may be able to claim a
right to indemnity (payment) for that liability from another party. This right to indemnity can be
based on law or contract. See, e.g., Marquez v. City Stores Co., 371 So. 2d 810 (La. 1979)
(finding no right to contractual or legal indemnity); 41 AM. JUR. 2D Indemnity § 2 (West 2014).
An example of a legal right to indemnity is the right of an employer to recover from its
employee, who has committed a tortious act for which he would be personally liable, where the
employer has been held vicariously liable for that tortious act of the employee. See, e.g.,
RESTATEMENT (SECOND) OF TORTS § 886B & cmt. e, at 344, 347 (1977); RESTATEMENT
(SECOND) OF AGENCY § 401 & cmts. a, e, at 237-38, 240 (1957); 27 AM. JUR. 2D Employment
Relationship § 408 (West 2014). As a general rule, claims of a right of indemnity are more
likely to be successful if there is a contractual basis for the claim. Compare, e.g., Hudson v.
Siemens Logistics & Assembly Sys., Inc., 353 F. App'x 717 (3d Cir. 2009) (accepting a claim
of contractual right to indemnity), with e.g., Williams v. Unit Handling Sys. Div. of Litton Sys.,
Inc., 449 N.W.2d 669 (Mich. 1989) (rejecting claim of right to implied indemnity against
employer for manufacturer's costs of settling suit for defect in a robotic machine with employee
of employer).
FLORIDA LAW REVIEW
which goods of that kind are used."4 6
Where personal injuries are
involved, the standard of "fit for ordinary purposes" requires that the
product be reasonably safe for such purposes.4 7 This implied-by-law
"warranty of merchantability" imposes a standard that is basically the
same as the tort standard used for products liability.48 Therefore, the
discussion below at Subsection II.A.2.b of the tort scheme for
determining defectiveness usually applies to UCC merchantability
claims.
Two UCC rules are especially important in considering the warranty
of merchantability. First, a seller/lessor can sometimes avoid being
subject to this implied warranty of merchantability by excluding or
modifying the warranty.49 Second, if the merchant does not exclude or
modify the warranty, the merchant: (1) will almost certainly be liable
for physical injuries caused by breach of the warranty, even though he is
entitled to limit or exclude liability for economic losses caused by the
breach,50 and (2) will likely be liable to third parties who are injured.
46. U.C.C. § 2-314 (1), (2)(c) (2012) (sales); U.C.C. § 2A-212(2)(c) (leases). A
manufacturer is treated as a merchant under these provisions because the manufacturer sells or
leases the product.
47. OWEN, supra note 44, at 32.
48. Hood v. Robi Am. Corp., 181 F.3d 608, 610 n.1 (4th Cir. 1999) (noting that, in the
case involved, the tort theories and the warranty theory were "virtually identical"); Denny v.
Ford Motor Co., 662 N.E.2d 730, 738 (N.Y. 1995) ("As a practical matter, the distinction
between the defect concepts in tort law and in implied warranty theory may have little or no
effect in most cases."); see RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2 cmt. n, at 35-36
(1997) ("Regardless of the doctrinal label attached to a particular claim, design and warning
claims rest on a risk-utility assessment."); id. at cmt. r, at 40 ("[I]n cases involving defectcaused
harm to persons or property, a well-coordinated body of law dealing with liability for
such harm arising out ofthe sale of defective products would adopt the tort definition of product
defect."); OWEN, supranote 44, at 32.
49. See U.C.C. §§ 2-316(2), 2A-214(2); see OwEN, supranote 44, § 4.9.
50. See U.C.C. § 2-302 ("If [a] court... finds the contract or any clause of the contract to
have been unconscionable ... the court may refuse to enforce the contract .... "); U.C.C. § 2-
719(3) ("Limitation of consequential damages for injury to the person in the case of consumer
goods is prima facie unconscionable but limitation of damages where the loss is commercial is
not."); U.C.C. § 2A- 108 (establishing a scheme for addressing unconscionability and authorizing
the court to grant appropriate relief); U.C.C. § 2A-503(3) ("Limitation, alteration, or exclusion of
consequential damages for injury to the person in the case of consumer goods is prima facie
unconscionable .... ); U.C.C. §2A-520(2) ("Consequential damages ... include ... injury to person
or property proximately resulting from any breach of warranty."). The definitions sections of
Articles 2 and 2A refer to U.C.C. § 9-109 for the definition of consumer goods. Section 9-102
defines consumer goods as goods "used or bought for use primarily for personal, family, or
house hold purposes." U.C.C. § 9-109(23).
51. See U.C.C. §§ 2-318, 2A-216. Both sections have three alternative provisions, which
vary in the breadth of coverage of third parties.
[Vol. 66
"SOPHISTICATED ROBOTS"
2. Tort
a. Basic Concepts
As indicated above, tort law operates as a corrective justice system
to return a victim to the position the victim would have occupied but for
the injury caused by defendant's wrongdoing. 52 A tort can be broadly
defined as a civil "wrong" (other than a breach of contract) that causes
injury, for which a victim can seek a judicial remedy, usually in the
form of damages. 53 Because this definition encompasses a broad range
of "wrongs," there is no single test or definition of wrong. Fortunately,
there is no need to consider all the meanings of "wrong" in this Article
because negligence generally serves as the basic test or principle to
identify wrongdoing where personal injury is involved.
Negligence is the most common form of "fault" in tort law and is
often defined by reference to a "reasonable person," whose conduct is,
by definition, never negligent.54 An alternative approach addresses
negligence in terms of a cost-benefit test based on a comparison of
accident costs with safety costs. 55 Under this approach, conduct is
negligent if a person, when deciding whether to engage in conduct
involving foreseeable risk of injury, chooses to engage in that conduct
without adopting feasible safety measures that would cost less than the
foreseeable injury. Defining the foreseeable injuries as accident costs
(AC) and the safety measures as safety costs (SC), the test can be stated
algebraically as follows: If AC > SC and an actor chooses to risk AC
rather than incur SC, the actor is negligent.56 The goal is to create
market incentives that internalize the costs of wrongdoing to the
wrongdoer. In this way, individual private actors decide the most
efficient way to address potential liability for wrongful injury.
Both tests rely on two basic terms or concepts: foreseeability and
feasibility. These terms focus on what could have been expected of a
person in the past at the time of the conduct involved in the eventual
wrongdoing. From the point of view of the time when the conduct was
undertaken, two questions arise: (1) What mishaps could the actor
reasonably foresee in the future; and (2) What actions to prevent
mishaps were reasonably feasible? Additional knowledge and new
52. See supranote 22 and accompanying text.
53. See, e.g., DOBBS, supra note 22, § 1, at 1-2; 1 FRANCIS HILLIARD, THE LAW OF TORTS
OR PRIVATE WRONGS 1-3 (1861); SIR FREDERICK POLLOCK, THE LAW OF TORTS: A TREATISE ON
THE PRINCIPLES OF OBLIGATIONS ARISING FROM CIVIL WRONGS IN THE COMMON LAW 3-4 (5th.
ed. 1897); WILLIAM L. PROSSER, HANDBOOK OF THE LAW OF TORTS 1-6 (4th ed. 1971).
54. RESTATEMENT (SECOND) OF TORTS § 283, at 12 (1977).
55. See id. § 291.
56. See, e.g., F. PATRICK HUBBARD & ROBERT L. FELIX, THE SOUTH CAROLINA LAW OF
TORTS 63-65 (4th ed. 2011).
FLORIDA LAW REVIEW
technology that was unavailable at the time of the conduct are excluded
from the assessment of negligence. Unfortunately, the terms
"reasonably foreseen" and "reasonably feasible" are, of necessity,
extremely vague.57 This Article addresses the approach to the use of
these vague terms in the context of specific applications.58
b. Products Liability: Sales, Leases, and Other Distributions
i. Doctrine
(a) The Three Types of Defects
In cases of sellers, lessors, and other distributors, tort law governs
liability for physical injury arising from products regardless of whether
contract law applies.5 9 Tort liability for product-caused injuries is based
57. But see generally Symposium, The Limits of Predictability and the Value of
Uncertainty,60 DEPAUL L. REv. 271 (2011) (discussing the benefits ofuncertainty in law).
58. See infra notes 62-76 and accompanying text.
59. RESTATEMENT (THiRD) OF TORTS: PRODS. LIAB. §§ 12-14, at 206, 221, 227 (1997).
Economic losses are treated differently from physical injury because contracts are viewed as a
more efficient method of allocation of economic losses arising from a product defect. As a
result, the tort system generally uses the "economic loss rule," which provides that, where there
is no physical injury or injury to property other than the product, only contract doctrine applies.
Section 21 ofthe Restatement ofProductsLiabilityprovides:
§ 21. Definition of "Harm to Persons or Property": Recovery for Economic
Loss
For purposes of this Restatement, harm to persons or property includes
economic loss if caused by harm to:
(a) the plaintiff's person; or
(b) the person of another when harm to the other interferes with an interest of
the plaintiff protected by tort law; or
(c) the plaintiff's property other than the defective product itself. Id. § 21, at
293.
Comment a to this section notes as follows:
Two major constraints on tort recovery give content to this Section. First,
products liability law lies at the boundary between tort and contract. Some
categories of loss, including those often referred to as "pure economic loss," are
more appropriately assigned to contract law and the remedies set forth in
Articles 2 and 2A of the Uniform Commercial Code. When the Code governs a
claim, its provisions regarding such issues as statutes of limitation, privity,
notice of claim, and disclaimer ordinarily govern the litigation. Second, some
forms of economic loss have traditionally been excluded from the realm of tort
law even when the plaintiff has no contractual remedy for a claim.
[Vol. 66
"SOPHISTICATED ROBOTS"
on the concept of "defect"--i.e., there is no liability unless a product
defect caused the injury. The legal system analyzes product defects in
terms of three specific types of defects: design, warnings and
instructions, and manufacturing.
In considering liability for product defects, it is important to keep in
mind that, while efficiency is the dominant value in allocating liability,
it is not the only value. The dominance of efficiency is reflected in the
general rule that a product manufacturer or distributor is not liable
unless there is wrong defined in terms of efficiency. Given this general
rule, it is understandable that tort law defines nearly all product defects
in negligence-like terms that require a plaintiff to show that certain
safety costs were not incurred by the defendant even though they were
cheaper than the accident costs that they would prevent. Given the need
to show a defect in this way, it is very likely that at least one person in
the chain of distribution is at "fault" in the negligence sense of the
term.60
Despite the importance of this emphasis on efficiency, products
liability contains pockets of "strict liability"-i.e., liability without
negligence. These pockets generally exist where tort doctrine's desire to
protect the reasonable expectations of humans injured by a product
"defect" can be furthered without undue "unfairness" to product
distributors.
6 1
Design Defects. A cost-benefit test like that used for negligence is
used for identifying a design defect. Under this test, a product "is
defective in design when the foreseeable risks of harm posed by the
product could have been reduced or avoided by the adoption of a
reasonable alternative design ... ,,62 A "reasonable alternative design"
is defined in terms of risk-utility or cost-benefit-i.e., the safer
alternative design's costs (in terms of such factors as manufacturing
costs, loss in utility of the product, and increase in risks of different
harms) must be less than the foreseeable injuries prevented by incurring
the costs of the safer alternative design. 63 For example, if it is feasible
Id.§ 21 cmt. a, at 293. The economic loss doctrine has been used to bar tort claims in numerous
cases involving computers and software. See, e.g., Transport Corp. of Am. v. Int'l Bus. Machs.
Corp., 30 F.3d 953, 955 (8th Cir. 1994) (holding that suit in tort for failure of disk drive was
barred by economic loss doctrine and that because seller had disclaimed all contract remedies,
buyer had no contract claim).
60. See infra notes 63-69 and accompanying text.
61. See infra notes 71-72, 80, 86 and accompanying text.
62. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2(b), at 14.
63. See id. at cmts. d, f, g, at 19, 22-23, 27; see also OWEN, supra note 44, at 508-32,
552-58. At one time, § 402A of the Restatement of the Law of Torts imposed liability on sellers
for injuries caused by "any product in a defective condition unreasonably dangerous to the user
or consumer or to his property." RESTATEMENT (SECOND) OF TORTS § 402A. The comments to
the section indicated that the determination of whether such a defect existed was based on
1821
FLORIDA LAW REVIEW
under the existing "state of the art" design64 to add to an automobile's
restraint system a shoulder harness that costs less than the costs of
accidents prevented by the shoulder harness, the harness should be
included in the system.65
If it is not, the system is defective, and the
manufacturer is liable for injuries caused by the defect.
Warning and Instruction Defects. The cost-benefit test is also
used for warnings and instructions. Thus, a product "is defective
because of inadequate instructions or warnings when the foreseeable
risks of harm posed by the product could have been reduced or avoided
by the provision of reasonable instructions or warnings by the seller or
other distributor .... ,66 These claims raise two issues: (1) was a
warning or instruction needed,67 and (2) if a warning or instruction was
provided, was that warning or instruction reasonable. As a general rule,
a warning or instruction concerning a serious risk is required if: (1) the
risk is foreseeable; and (2) a significant number of users will not be
aware of the risk.68
Determining whether "reasonable instructions or warnings" were
used involves a contextual consideration of "various factors, such as
"consumer expectations" and "cost-benefit" analysis. See id., cmts. g, h, i, k; F. Patrick
Hubbard, Efficiency, Expectation, and Justice: A JurisprudentialAnalysis of the Concept of
UnreasonablyDangerousProductDefect, 28 S.C. L. REV. 587, 604-21 (1977); David G. Owen,
Bending Nature, Bending Law, 62 FLA. L. REV. 569, 580-606 (2010). The "consumer
expectations" test presented so many problems that it is "withering" as a free-standing basic test.
OWEN, supra note 44, at 310; see infra notes 321-23 and accompanying text. Currently, the test
is only used in a small "dwindling" number of states. OWEN, supra 44, at 310. In the other
states, consumer expectations are just a factor to consider in applying the cost-benefit test or as
the basis for a few narrow pockets of strict liability. Id. For discussion of instances of strict
liability, see infra notes 71-72, 78-86.
64. For discussion of "state of the art," which can apply to both foreseeability of risk and
feasibility of safety measures, see, for example, RESTATEMENT (THIRD) OF TORTS: PRODS. LInB.
§ 2 cmt. d, case law and commentary IV-B, at 80-81 (noting the role of "good sense" and
"pragmatism" in applying the concept); OWEN, supra note 44, § 10.4, at 706-07 (discussing
state of the art and noting that, though the concept is "unrefined" and evolving, it reflects the
"reluctance to impose liability ... for dangers that were unknowable, or unpreventable").
65. See, e.g., Williamson v. Mazda Motor of Am., Inc., 131 S. Ct. 1131, 1134 (2011)
(holding that claim based on lack of shoulder belt was not preempted). The costs of the
harnesses per vehicle must be multiplied by the number of vehicles manufactured to get the total
safety costs. The total injury costs include all foreseeable injuries, each of which must be
discounted by the probability that the injury will actually occur.
66. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2(c).
67. For example, an ordinary paring knife does not need a warning that it is sharp.
68. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 10; HUBBARD & FELIX, supra note
56, at 329. T-h-re are "exceptions" to this rule. See, e.g., RESTATEMENT (THIRD) OF TORTS:
PRODS. LIAe §6(d) (discussing mandatory instructions and warnings for prescription drugs and
medical devices); OWEN, supra note 44, § 9.5, at 620-24 (explaining that under the
"sophisticated users doctrine," there is no need to warn because of the expertise of the buyer and
under the "bulk suppliers doctrine," there is no duty to warn if there is no way to warn because
of the nature of the product). In limited circumstances, a distributor may have a post-sale duty to
warn. See infranotes 251-52 and accompanying text.
[Vol. 66
"SOPHISTICATEDROBOTS"
content and comprehensibility, intensity of expression, and the
characteristics of expected user groups."69 There is no "easy guideline,"
and "the ability of a plaintiff to imagine a hypothetical better warning in
the aftermath of an accident does not establish that the warning actually
accompanying the product was inadequate." 70
Manufacturing Defects. A manufacturing defect results when a
product is not manufactured in accordance with the manufacturer's
specifications. For example, the product would have a manufacturing
defect ifthe blueprint specified four bolts for an assembly but only three
bolts went into that particular assembly for the product. This approach,
in effect, replaces the cost-benefit test used for design and warnings
with a "blueprint test" that relies on the manufacturer's own
specifications. This test results in a form of strict liability because a
product will be held to be "defective" even if the manufacturer used a
cost-effective method to control quality in the manufacturing process. 71
The legal system views this approach as fair because: (1) unlike design
and warning defects, the entire product line is not defective; (2) the
product is, in fact, defective when measured by the manufacturer's own
design specifications; (3) consumers (and third parties affected by the
defect) are entitled to expect compliance with product specifications; (4)
nonmanufacturing sellers may be able to pass liability costs up the
distribution chain to the manufacturer; and (5) the manufacturers (and
other sellers) are better able than victims to distribute the cost of the
72
injuries.
(b) Foreseeability, Misuse, and Obvious Risks
Two types of problems have been particularly troublesome in terms
of both design and warning: "misuse" and "obvious" risks. The broad
concept of "foreseeable use" addresses both. For example, if it is
69. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB., § 2(c) cmt. i, at 29; see also OWEN,
supra note 44, §§ 9.3-9.4, at 593-619 (addressing reasonableness in terms of "adequacy" and
"persons to be warned").
70. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2(c) cmt. i, at 29.
71. Id. § 2(a), at 14 ("[A] manufacturing defect [exists] when the product departs from its
intended design even though all possible care was exercised in the preparation and marketing of
the product ....");see, e.g., OWEN, supra note 44, §§ 7.1-7.4, at 446-75 (addressing strict
liability for manufacturing defects); Jurls v. Ford Motor Co., 752 So. 2d 260, 266 (La. Ct. App.
2000) (holding that plaintiff had presented sufficient circumstantial evidence of manufacturing
defect to withstand motion for directed verdict).
72. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 1 cmt. a, at 5-6; id.§ 2 cmt. a, at
14-15; § 2 rptr. n. to cmt. d, at 70; see, e.g., F. Patrick Hubbard, Reasonable Human
Expectations: A Normative Model for Imposing Strict Liability for Defective Products, 29
MERCER L. REv. 465 (1978) (arguing for protection of reasonable human expectations as a value
that competes with efficiency and justifies imposing liability on profit-oriented business entities
in the chain of distribution regardless of whether that entity can be charged with efficiencydefined
"fault").
FLORIDA LAW REVIEW
foreseeable that a product will be used in a manner other than its
intended use, design and warning decisions must be based on that
foreseeability.73 As a result, because a drunken driver might foreseeably
misuse automobiles, they must be "crashworthy,, 74 which means they
must have cost-effective safety measures, such as doors that will not
pop open in a collision regardless of whether the driver was
intoxicated.75 Similarly, even though a dangerous risk may be
"obvious," the risk must be addressed by effective warnings or a design
approach if it is foreseeable that the risk will not be noticed or
appreciated or that the user will not use due care to avoid the risk.76
(c) Strict Liability
One basic rule of products liability is that sellers are not liable where
designs or warnings/instructions achieve an efficient balance of accident
costs and safety costs. This rule imposes the costs of efficient injuries
on innocent victims. Partly because of this imposition, a "strict liability"
approach replaces this efficiency-oriented test on grounds of fairness to
victims in a number of instances.77 As indicated above, manufacturing
defects are one such instance. This Subsection addresses two other
instances of strict liability.7 8
73. See RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2 cmts. f, m, p, at 22-23, 33-34,
38-39; OWEN, supra note 44, § 13.5, at 890-95 (addressing the "Foreseeability Limitation"),
§ 17.3, at 1131-46 (addressing "[c]rashworthiness"). Where a seller is liable as a result of
foreseeable misuse or other misconduct, the person engaging in the misuse or other misconduct
(including the victim if the victim has engaged in such conduct) may also be liable to some
extent as well. See RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2 cmt. p, 38-39; § 17;
OWEN, supranote 44, § 13.5, at 896-98 (addressing comparative fault). The application of this
principle can be very complex. See, e.g., Randy R. Koenders, Annotation, ProductsLiability:
Liability of Manufactureror Seller as Affected by Failureof SubsequentParty in Distribution
Chain to Remedy or Warn Against Defect of Which He Knew, 45 A.L.R.4th 777 (1986)
(discussing state and federal products liability cases).
74. See RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 16 cmt. a, at 236; OWEN, supra
note 44, § 17.3, at 1131-46.
75. See Daly v. Gen. Motors Corp., 575 P.2d 1162, 1173-74 (Cal. 1978) (holding that
evidence of intoxication in crashworthy case must be excluded or its effect confined to relevant
issues so that it is not misinterpreted by the jury as totally barring recovery for lack of
reasonable crashworthiness).
76. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB., § 2 cmts. j, 1,at 31, 33; see also
OWEN, supranote 44, § 10.2, at 647. Where it is not foreseeable that an obvious danger will not
be noticed, there is not duty to warn. See, e.g., Jones v. W+M Automation, Inc., 818 N.Y.S.2d
396, 399 (N.Y. App. Div. 2006) ("[T]here is no duty to warn of an open and obvious danger of
which the product user is actually aware or should be aware as a result of ordinary observation
or as a matter of common sense.").
77. See supranotes 71-72 and accompanying text.
78. Though not common, additional instances can also arise. See, e.g., RESTATEMENT
(THIRD) OF TORTS: PRODS. LAB. §§ 12-14, at 206-30 (addressing the liability of successors and
apparent manufacturers).
[Vol. 66
"SOPHISTICATED ROBOTS"
First, strict liability results from the imposition of liability for defects
on all product distributors. As indicated in the discussion of design and
warning defects, the use of the cost-benefit approach to identify design
defects requires that liability be based on at least one product
distributor's fault. However, if a product defect causes injury, a person
in the business of selling or otherwise distributing the product is
"strictly liable" for that injury regardless of whether that person was not
"at fault" in the sense of being the person who made the erroneous costbenefit
decision.79 For example, a manufacturer that did not include a
particular type of shoulder harness as a part of a vehicle's safety system
when it would have reduced accidental injuries in a cost effective
manner was negligent and thus "at fault." In contrast, wholesalers and
retailers who sell the automobile are not at fault. Nevertheless, they are
legally liable as sellers of a defective product. Consequently, it is
79. See RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 1, at 5 (applying scheme to
commercial sellers and lessors); id. § 19, at 267 (defining "product" as "tangible personal
property distributed commercially for use or consumption"); id §20, at 284 (defining "seller"
and "distributor"). The concept of "selling or otherwise distributing" includes sellers of
components if:
(a) the component is defective in itself, as defined in this Chapter, and the
defect causes the harm; or
(b)(1) the seller or distributor of the component substantially participates in the
integration of the component into the design of the product; and
(2) the integration of the component causes the product to be defective, as
defined in this Chapter; and
(3) the defect in the product causes the harm.
Id. §5, at 130. Special rules are established for particular types of "sellers." See, e.g., id. §6, at
144 (discussing liability for sellers of prescription drugs and medical devices); id.§ 7, at 160
(discussing liability for sellers of food products); id.§ 8, at 166 (discussing liability for sellers of
defective used products); id. §§ 12-13, at 206, 221 (discussing liability for successors of a
business that has previously sold a defective product and failed to warn consumers post-sale);
id. § 14, at 227 (discussing liability for sellers of products sold as if they manufactured the
product even though it was manufactured by another). Licensors can, under many
circumstances, be subject to the same liability as the seller of a product. See, e.g., OWEN, supra
note 44, § 16.2, at 1070-77.
80. See supra note 79 and accompanying text. The Restatement of Torts: Products
Liability notes that its approach results in liability without fault. See RESTATEMENT (THIRD) OF
TORTS: PRODS. LAB. § 1 cmts. e, o, at 8-9, 38. The Restatement (Second) of Torts, which
follows a similar approach, states:
The basis for the rule is the ancient one of the special responsibility for the
safety of the public undertaken by one who enters into the business of
supplying human beings with products which may endanger the safety of their
persons and property, and the forced reliance upon that undertaking on the part
ofthose who purchase such goods.
FLORIDA LAW REVIEW
common to say that they are "strictly liable."81 Similarly, all distributors
are strictly liable for manufacturing defects.8 2
The reasons for this treatment are, in part, similar to the reasons for
imposing strict liability for manufacturing defects.83 In addition, the
imposition of liability on all distributors is based on a concern for
insuring that the person injured as a result of the defect can find a
solvent party able to provide compensation adequate to provide
corrective justice.
84
Second, strict liability results from the following rule: a distributor
who makes a non-negligent, non-fraudulent material misrepresentation
of fact concerning the product is liable for injury caused by that
misrepresentation even if the misrepresentation was innocent.8 This
instance of strict liability is similar to the liability imposed for
manufacturing defects because the misrepresentation, like the product
specifications, provides a standard for defining defect. Thus, the reasons
for strict liability in this instance parallel the reasons for strict liability
for manufacturing defects.86
ii. Practice
Specialization. To a considerable extent, products liability law is a
specialized practice area for both defendants' and plaintiffs' attorneys.
This specialization results partly because the area involves a distinct set
of doctrinal rules. The American Law Institute's choice to adopt the
Restatement of Torts: Products Liability as a separate publication
reflects this doctrinal distinctiveness.
Plaintiffs' attorneys. Allegations that a product is defective in terms
of design or warning generally require an expert qualified to testify
about reasonable alternative designs or warnings. The plaintiff's
RESTATEMENT (SECOND) OF TORTS, supra note 45, § 402A cmt. f, at 350-51.
81. Technically, it might be more accurate to view this approach as a form of vicarious
liability for the wrong of another in the way, for example, employers are vicariously liable for
the wrongs of their employees. See supranote 45. However, at this point in time, the term "strict
liability" is so widely used that a change in terms is not likely.
82. See supra notes 71-72 and accompanying text.
83. See supra note 72 and accompanying text.
84. See RESTATEMENT (THIRD) OF TORTS: PRODS. LiAB.§ 1 cmt. e, at 8-9.
85. Id. § 9, at 187. A similar result can apply to an express warranty. See supra note 44
and accompanying text.
86. Compare id.§ 9 cmt. b, at 187-88, with supranote 72 and accompanying text.
87. See RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2 cmt. f, at 22-24; see also
OwEN, supranote 44, § 6.3, at 362-90 (addressing the use of experts to prove whether a defect
exists). An expert is not required where the feasibility of a reasonable alternative design is
obvious to and understandable by laypersons. The Restatement of Torts: Products Liability
notes:
[W]hen a manufacturer sells a soft stuffed toy with hard plastic buttons that are
[Vol. 66
"SOPHISTICATED ROBOTS"
attorney pays for the expert's services as the case is being prepared and,
if necessary, tried. Moreover, as indicated below in the discussion of the
litigation process, product liability suits demand a considerable
investment of the attorney's time and money. Under the contingency fee
system, the attorneys do not get any return on their investment unless
the case resolves-by trial or settlement-in the plaintiffs favor.
Because this process can take years, these attorneys must, to some
extent, have "deep pockets." The net effect of these economic
considerations is that plaintiffs' attorneys will not accept or bring a suit
unless the injuries are severe enough to justify an amount of damages
sufficient to make the suit worthwhile.
As a result, product sellers and lessors are not likely to be held liable
for defects that do not cause substantial personal injuries because such
suits are not worth bringing. On the other hand, ifthe plaintiff's injuries
are so severe that a very substantial amount of damages is involved, a
plaintiff's attorney may take a case even if the odds are against success.
Like any entrepreneur, it is rational for a plaintiffs' attorney to invest in
a project with a less than fifty percent chance of success if the costs of
the project are less than the likely percentage of return multiplied by the
likely amount ofreturn.
Also, like any entrepreneur, a plaintiffs attorney will seek new
business. The primary way to do this is to get new clients. Thus, it is not
uncommon to see attorneys engaging in advertising that informs
potential plaintiffs of the possibility of suit if they have been injured by
a particular product.
Defendants' attorneys. Because of the utility of products liability
insurance, 88 insurance companies usually hire defense attorneys, who
are frequently paid on a fee basis. In most cases, insurance companies
will decide the important issues such as whether to make or accept a
particular settlement offer.89
Experts. The experts required in a particular products case depend
on the circumstances. For robotic products, an expert would be required
to have sufficient engineering expertise to provide a reliable opinion on,
easily removable and likely to choke and suffocate a small child who
foreseeably attempts to swallow them, the plaintiff should be able to reach the
trier of fact with a claim that buttons on such a toy should be an integral part of
the toy's fabric itself (or otherwise be unremovable by an infant) without hiring
an expert to demonstrate the feasibility of an alternative safer design.
Furthermore, other products already available on the market may serve the
same or very similar function at lower risk and at comparable cost. Such
products may serve as reasonable alternatives to the product in question.
RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 2 cmt. f, at 22-24.
88. See supranotes 38-41 and accompanying text.
89. The defendant client has rights in this process; however, partly because of typical
contract terms, the insurer often has considerable power over litigation decisions.
1827
for example, the existence of a reasonable alternative design.90 In terms
of both design and warning, a "human factors expert," who might have
expertise in a field like psychology, could also testify.91
Process: discovery, summary judgment, settlement, trial, and
appeal. The resolution of a products liability case generally takes a
considerable amount of time. Well before trial, each side will engage in
discovery on the other side's experts by, for example, requesting the
production of documents and reports and by deposing-taking sworn
testimony from-one another's experts as well as other potential
witnesses. This part of the process can easily take more than a year.
Typically, after most discovery is completed, a defendant will file a
pretrial motion-often termed "Motion for a Judgment as a Matter of
Law" or "Motion for Summary Judgment"-for a determination of
whether the plaintiff has produced sufficient evidence to support a
finding by a "reasonable jury" that the product was defective and that
the defect caused the injury.92 The court will grant this motion if, for
example, the court finds that the plaintiffs expert lacks sufficient
expertise or if, at trial, the evidence developed in discovery cannot
support the expert's opinion.93 If the motion is granted, judgment is
entered in favor of the defendant, and the case will not reach the jury.
If the motion is denied, the parties are likely to settle the case instead
90. See, e.g., Hills v. Fanuc Robotics Am., Inc., No. 04-2659, 2010 WL 890223, at *5
(E.D. La. Mar. 5, 2010) (holding that a licensed mechanical engineer was qualified to testify
about the defect in a robot used to stack crates on wooden pallets even though he was not a
specialist in robotics); cf, e.g., Knight v. Otis Elevator Co., 596 F.2d 84, 87 (3d Cir. 1979)
(finding that the expert's lack of background in design and manufacture of elevators went to the
permissible scope of expert's testimony, not admissibility); Murphy v. Montgomery Elevator
Co., 957 S.W.2d 297, 299 (Ky. Ct. App. 1997) (concluding that a mechanical engineer was
qualified to testify concerning the design of and warnings for an escalator and that his lack of
background in the field attacked the expert's credibility, not the admissibility of his testimony);
Graves v. CAS Medical Sys. Inc., 735 S.E.2d 650, 655-59 (S.C. 2012) (affirming grant of
summary judgment on ground that testimony of software experts concerning failure of medical
monitoring system was inadmissible).
91. For discussion of human factors in robotic design, see infra notes 253-54 and
accompanying text. See also, e.g., Humphries v. Mack Trucks, Inc., 198 F.3d 236, at *4 (4th
Cir. 1999) (holding expert testimony by psychologist was properly admitted); Lucas v. Dorsey
Corp., 609 N.E.2d 1191, 1197 (Ind. Ct. App. 1993) (opinion of human factors expert sufficient
to support denial of motion for summary judgment).
92. See, e.g., FED. R. Civ. P. 56.
93. See, e.g., supra note 87 and accompanying text and infra notes 155-66 and
accompanying text. But cf OWEN, supra note 44, § 6.3, at 366 (stating that in some cases, juries
can comprehend the facts ofa defective design case without the use of expert testimony). Expert
testimony may also be required to show that the defect caused the plaintiff's injury. See OWEN,
supra note 44, § 6.3, at 367. Where the expert is qualified, his opinion may be enough to support
a denial of the motion for summary judgment. See, e.g., Lucas, 609 N.E.2d at 1199-1201
(holding the testimony of a human factors expert sufficient to support the denial ofa motion for
summary judgment). Occasionally, no expert witness on defect will be required. See supranote
87.
1828 FLORIDA LA W REVIEW [Vol. 6
"SOPHISTICATEDROBOTS"
of trying it because: (1) the results of a jury trial are hard for either side
to predict; and (2) a large amount of money is likely at issue since, as
indicated above, incentives drive plaintiffs' attorneys to bring only
those claims that have sufficient damages to justify their time and
expense. 94
Thus, as settlement negotiations proceed, the plaintiffs
attorney is concerned about the possibility of no recovery or a totally
inadequate recovery and the defendant's attorney is worried about the
potential liability for a large full recovery of damages.
If the defendant's pretrial motion is granted, the plaintiff is likely to
appeal unless the defendant makes a reasonably high offer to settle.
Absent such an offer, an appeal is likely because the marginal cost to
the plaintiff of an appeal (beyond the costs already expended) is
relatively low compared to the possible gain by a reversal on appeal or a
settlement pending appeal, given the likelihood that the case would not
have been brought unless a substantial amount of damages is involved.
If the appeal is successful, the case may settle or return for trial.
c. Reasonable Care for Control, Use, and Maintenance
Tort law governs most liability for physical injury caused by persons
who control, use, or maintain machines. These persons are generally
liable if such injuries were caused by their negligence-i.e., a failure to
exercise a reasonable level of care95-in terms of: (1) use of the
machine;96 (2) maintenance (or choice of a person to perform
maintenance) of the machine;97 (3) supervision or authorization of the
machine's use;98
or (4) prevention of others from using the machine if
the machine is foreseeably dangerous.99
In this context, reasonable care is often articulated in terms of the
conduct of a "reasonable person" rather than in the explicit cost-benefit
94. See supratext accompanying notes 87, 90-91; infranote 270 and accompanying text.
95. See supranotes 54-58 and accompanying text.
96. For example, one is liable for negligent driving of an automobile. See infra note 100.
Ownership of a car also involves a duty of due care in maintenance. See, e.g., 57A AM. JUR. 2D
Negligence §§ 329-341 (West 2014).
97. See, e.g., 38 AM. JUR. 2D Garages,Service Stations, andParkingFacilities§ 53 (West
2014) (addressing liability for negligent repairs); 57A AM. JUR. 2D Negligence §§ 310-325
(addressing liability for negligent entrustment); id. § 367 (addressing liability for negligent
maintenance).
98. See, e.g., Mohler v. St. Luke's Med. Ctr., LP, No. 1 CA-CV 08-0078, 2008 WL
5384214, at 16 (Ariz. Ct. App. 2008) (holding that jury issue existed as to whether hospital
negligently credentialed physician for use of surgical robot); RESTATEMENT (SECOND) OF TORTS
§ 308, at 100 (1977) (discussing negligence of permitting improper persons to use things or
engage in activities).
99. See Jupin v. Kask, 849 N.E.2d 829, 832 (Mass. 2006) (discussing the duty of due care
to secure a handgun); 57A AM. JUR. 2D Negligence §§ 299, 327-28 (West 2014) (addressing
requirement of reasonable precautions to prevent injury from foreseeable acts of third parties,
including children).
1829
FLORIDA LA W REVIEW
terms addressed above.' 00 Both tests foster an efficient level of
reduction in injury costs. The primary difference is that the explicit
costs-benefit phrasing better captures the complex, expertise-driven
tasks of designing products and articulating warnings,10 while
reasonable-person phrasing is easier for a jury to apply because a
layperson can understand the activity involved.1 0 2 Where laypersons
lack adequate understanding of an activity-for example, in evaluating
the conduct of a doctor performing an operation with a robotic surgical
system-the testimony of a qualified expert would be required to show
the standard of reasonable conduct.' 03
i. Examples
(a) Employment
Because so many robots are used in factories and other employment
settings,1°4
injuries to workers by robotic machines have been the
impetus for employees to bring numerous claims. 10 5 Tort suits against
employers are complicated by the fact that employees injured on the job
are usually covered by workers' compensation. 0 6
In most states,
workers who are injured while working for their employer cannot sue
the employer in tort; workers' compensation is their exclusive
remedy, 07 In effect, the employer is immune from most tort suits
100. See supra notes 56, 62-63 and accompanying text. For example, the Massachusetts
Supreme Court expressed the standard in an early automobile case, stating that "[t]he law as to
drivers of motor vehicles is not different from that which governs other persons. The standard
required is that ofthe reasonably prudent person under all the circumstances." Massie v. Barker,
113 N.E. 199, 199 (Mass. 1916).
101. See HUBBARD & FELIX, supra note 56, at 63-65 (discussing generally the use of the
"calculus ofrisk" balancing test).
102. See id. at 65 (discussing generally the use of the "reasonable person" approach to
negligence).
103. See id. at 146-49 (discussing the standard of care for professional malpractice). An
expert might also be required to show breach of the standard and injury from that breach. Id.
104. See infranote 153 and accompanying text.
105. See infra notes 155-56 and accompanying text for discussion of products liability
claims for industrial robots.
106. See I ARTHUR K. LARSON & LEX LARSON, LARSON'S WORKER'S COMPENSATION LAW
§ 2.08, available at https://advance.lexis.com/GoToContentView?requestid=168cfb53-59db-
4f29-b63f-581d6Oc69eOc&crid=f681d21f-574f-5628-3461-8b2f8ea22c56 (paid subscription
required) (last visited June 25, 2014); see, e.g., Delawder v. Am. Woodmark Corp., 178 F.
App'x 197 (4th Cir. 2006) (holding that suit against employer for injury from robotic paint
machine barred by exclusivity doctrine); Owens v. Water Gremlin Co., 605 N.W.2d 733 (Minn.
2000) (holding that the family of a worker who suffered permanent partial disability and
eventually died from being pinned against a wall by a robot arm could recover worker's
compensation benefits).
107. Delawder, 178 F. App'x at 199; LARSON & LARSON, supra note 106, § 100.01. An
employee injured on the job is not usually barred from suing a third party-for example, the
[Vol. 66
"SOPHISTICATED ROBOTS"
brought by an employee. This immunity is based on a principle of
fairness: Workers receive guaranteed no-fault workers' compensation
for injuries while working; in exchange, workers give up the right to sue
in tort, where recovery might be higher but is far less certain.108
Exceptions to this immunity exist in most states. For example, some
states have an exception where the employer acted with intent to injure
or with a "deliberate intention" of exposing the worker to a specific
unsafe working condition. Workers have made claims against their
employers for injuries from robots based on such exceptions, but these
claims have had only limited success. 1
0
9
(b) Premises Liability
As a general rule, persons with control over premises have a duty to
make the premises reasonably safe for persons coming on the premises
to transact business. 10 Early robotic machines, such as automated
elevators and escalators, have been the subject of premises liability
claims, and many plaintiffs have sued for injuries allegedly cause by
lack of reasonable care in the operation and maintenance of these
automated machines."' In recent years, cases have also involved claims
manufacturer of a workplace machine. Workers have relied on this rule to sue manufacturers of
robotic machines. See infranote 155 and accompanying text.
108. LARSON & LARSON, supranote 106, § 100.01[1].
109. Compare, e.g., Delawder, 178 F. App'x at 202 (holding that the exception was not
proved and that, therefore, the exclusivity doctrine barred suit), Miller v. Rubbermaid, Inc., No.
23466, 2007 WL 1695109 (Ohio Ct. App. June 13, 2007) (finding that there was insufficient
evidence of intentional conduct by employer), and Edens v. Bellini, 597 S.E.2d 863, 872 (S.C.
Ct. App. 2004) (finding that there was insufficient evidence to show commission of an
intentional tort), with e.g., State ex rel. Scott Fetzer Co., Halex Div. v. Indus. Comm'n of Ohio,
692 N.E.2d 195 (Oh. 1998) (finding an intentional violation by an employer), and Behurst v.
Crown Cork & Seal USA, Inc., No. 04-1261-HA, 2007 WL 987452, *7 (D. Or. Mar. 30, 2007)
(holding that a jury could find that there was sufficient evidence to support intentional conduct
by an employer).
110. RESTATEMENT (SECOND) OF TORTS § 332 & cmt. a (discussing "invitee"); id.
§§ 341A-344, at 209-10 (listing duties owed to invitee).
111. See, e.g., Estabrook v. J.C. Penny Co., 464 P.2d 325 (Ariz. 1970) (en banc) (affirming
grant of new trial to owner in action resulting from injuries sustained by a minor while playing
on an escalator); Giles v. City of New Haven, 619 A.2d 476 (Conn. App. Ct. 1993), aff'd, 636
A.2d 1335 (1994) (reversing jury verdict for defendant owner in negligence suit for failure to
inspect, maintain, and repair an elevator); Brown v. Sears, Roebuck & Co., 514 So. 2d 439 (La.
1987) (affirming verdict in favor of plaintiff in escalator case where infant sustained injuries
after his finger was caught in the space between the moving treads and a panel); Guilfore v.
D.H. Holmes Co., 631 So. 2d 491 (La. Ct. App. 1994) (reversing judgment against store in
defective elevator suit); M & R Inv. Co. v. Anzalotti, 773 P.2d 729 (Nev. 1989) (reversing grant
of plaintiffs motion for new trial and reinstating jury verdict for defendant owner in negligence
action against elevator owner); Phillip White, Jr., Annotation, Liability ofMaintainer,Repairer,
or Installerof Automated PassengerElevatorfor Injury Resulting from Use of Elevator, 115
A.L.R.5th 1 (2004) (discussing federal and state cases on liability of maintainer, repairer, or
FLORIDA LA WREVIEW
by factory visitors claiming they were injured by a failure to use
reasonable care in inspecting and operating robotic machines.l12
One group of cases involves suits against landlords, who have an
obligation to use reasonable care in providing security from violent
crime in a common area such as a lobby or a hallway.' 3 Some cases
have held that, where a landlord had humans providing security,
replacing them with an interlock and buzzer system did not satisfy the
landlord's duty because these were less effective than humans. 1 4
Though cases like these may suggest that the robotic security
systems should be measured by the capabilities of humans,' 15 two points
indicate that the holdings of the cases are not that broad. First, they
involve the replacement of humans rather than a situation where human
security had not been provided. Second, liability is based on the
standard of reasonable care. 1 6 Consequently, the only issues would be:
(1) whether, given the foreseeable risk of criminal attack, security
measures were necessary; and (2) if security measures were necessary,
whether the particular measure-whether human, robotic, or
otherwise-was reasonable." 7 Thus, for example, under the costbenefit
test, humans would not be required for security at all iftheir cost
as a safety measure exceeded the cost of foreseeable injuries from lack
of human security. On the other hand, if the costs of human security
were less than the foreseeable injuries from attacks, humans would be
necessary unless robots could provide at least a similar level of security.
installer of elevator); Philip White, Jr., Annotation, Liability of Building Owner, Lessee, or
Manager for Injury or Death Resulting from Use of Automatic Passenger Elevator, 99
A.L.R.5th 141 (2002) (discussing federal and state cases on the liability of owner, lessee, or
manager of building with automatic elevator); infranote 147 and accompanying text.
112. See, e.g., Rodriguez v. Brooks Pari-Automation, Inc., No. Civ.A.3:03CV00515-L,
2003 WL 21517851, at *1 (N.D. Tex. June 30, 2003) (alleging injuries from accident that
occurred while plaintiff was installing a robot with the defendant); Budris v. Robotic Res., R2,
Inc., 1997 WL 408717, at *1 (Conn. Super. Ct. July 11, 1997), modified, No. CV 91036468,
1998 WL 46224 (Conn. Super. Ct. Jan. 22, 1998) (claiming injuries from defendant's testing of
a robot while plaintiff electrician was working at defendant's premises).
113. See DOBBS, supranote 22, § 325, at 880.
114. See, e.g., Kline v. 1500 Mass. Ave. Apt. Corp., 439 F.2d 477, 486 (D.C. Cir. 1970)
(holding that where an apartment complex removed humans performing security functions, "the
same relative degree of security should have been maintained" and noting that it was not
deciding whether a "tenant-controlled intercom-automatic latch system.. . in the common
entryways" would have been sufficient); Green Cos. v. Divincenzo, 432 So. 2d 86, 87-88 (Fla.
Dist. Ct. App. 1983) (affirming that the landlord had a duty to institute procedures to keep
premises safe).
115. See infranotes 262-64 and accompanying text.
116. See supranotes 54-58, 101-03 and accompanying text.
117. See, e.g., DOBBS, supra note 22, § 325. The same analysis applies to the duty of a
business to its customers. Thus, for example, guards are not required for a store unless the
guards are a cost-effective measure to prevent crime in the parking lot. Posecai v. Wal-Mart
Stores, Inc., 752 So. 2d 762, 768 (La. 1999).
[Vol. 66
"SOPHISTICATED ROBOTS"
Similarly, the need for and adequacy of robotic security would depend
on the relative costs of foreseeable injuries compared to the costs of the
robotic security measures that might be used.
ii. Exception for Abnormally Dangerous Activities
An exception to the general rules of negligence applies in the case of
"abnormally dangerous activities." The "rule" concerning such
activities is easy to state, but very difficult to apply. Section 20 of the
ThirdRestatement provides:
(a) An actor who carries on an abnormally dangerous
activity is subject to strict liability for physical harm
resulting from the activity.
(b) An activity is abnormally dangerous if:
(1) the activity creates a foreseeable and highly
significant risk of physical harm even when
reasonable care is exercised by all actors; and
(2) the activity is not one of common usage. 118
Except for clear cases such as activities involving explosives and
wild animals, 119 determining whether a risk of physical harm is
"foreseeable" and "significant" even "when reasonable care is
exercised" and whether an activity is "one of common usage" has
challenged courts for decades.12
As a result of this problem, as well as
the current lack of knowledge about the specific characteristics of
sophisticated robots, it is virtually impossible to do more than guess
how a judge might treat a claim that using a sophisticated robot in a
particular way constituted an abnormally dangerous activity.'21
The lack of certainty concerning the application of the abnormally
dangerous activity doctrine could inhibit innovation because the
potential increased liability could have two effects. First, the expanded
potential liability of innovators could negatively affect their decision to
develop, for example, an Unmanned Aircraft System (UAS). Second,
118. RESTATEMENT (THIRD) OF TORTS: LIAB. FOR PHYSICAL AND EMOTIONAL HARM § 20, at
229 (2005).
119. See, e.g., id. § 20 cmt. e, at 233-34 (discussing blasting as "paradigm of an
abnormally dangerous activity"); id. § 22, at 293 (discussing strict liability for wild animals);
DOBBS, supra note 22, § 348, at 954 (discussing strict liability for explosives and high-energy
activities).
120. See DOBBS, supranote 22, §§ 346-51.
121. The determination of whether an activity is abnormally dangerous is a matter of law
for a judge to determine, not the jury. RESTATEMENT (THIRD) OF TORTS: LIAB. FOR PHYSICAL AND
EMOTIONAL HARM § 20 cmt. 1,at 240-41.
FLORIDA LAW REVIEW
the increased liability could also reduce the demand for a robotic
vehicle like a UAS because purchasers and users would need to worry
about potential greater liability for personal injury.
Three points provide guidance in "guessing" whether use of a
robotic machine is abnormall dangerous. First, today's automobiles are
not abnormally dangerous. 2 Second, the status of aircraft is unclear,
but recent cases have tended to view them as not abnormally
dangerous. 123
Third, courts might consider an experimental driverless
vehicle or UAS to be abnormally dangerous because experimenting
with large, mobile, sophisticated robots is not currently common and
arguably creates a "foreseeable and highll' significant risk of harm even
when reasonable care is exercised." 12 This third point does not
necessarily mean that increasingly sophisticated vehicles are
"experimental." There are large, important differences between a
"Google car" and a mass-produced Volvo with robotic features such as
electronic stability control and blind spot indicators. For example, the
Volvo's design is based on prior experimentation and testing and is
subject to regulation by the National Highway and Safety
Administration and is similar to automobiles sold by other
manufacturers. 25
d. The Role of Standards
Where diverse companies manufacture mass-produced goods such as
automobiles, some degree of standardization is necessary so that, for
example, drivers can drive many different cars with a minimal learning
curve, other drivers and pedestrians will know how the vehicles around
them respond, and parking lots and garages can be designed efficiently.
Where safety standards concerning the design and use of products are
involved, standardization is sometimes achieved informally by custom.
However, because of the advantages in terms of safety and
predictability of carefully prepared written standards, a more formal
approach has become increasingly common.1 26 Formal written
standards, including standards for industrial robots, 127 are often
developed by independent private entities like the Society of
122. See id.§ 20 cmt. j, at 238-39 (stating that an activity that in is in common usage is not
abnormally dangerous, even if it involves significant risk when reasonable care is exercised, and
discussing the use of automobiles as an example).
123. DOBBS, supranote 22, § 349, at 958.
124. See, e.g., Green v. Ensign-Bickford Co., 595 A.2d 1383, 1386-88 (Conn. App. Ct.
1991) (holding experimental use of volatile chemicals constituted an abnormally dangerous
activity).
125. See infranotes 134-35 and accompanying text.
126. See, e.g., OwEN, supranote 44, § 2.3, at 83-84.
127. See infra note 132 and accompanying text. For discussion of the definition and
number of industrial robots, see infra note 153.
[Vol. 66
"SOPHISTICATEDROBOTS"
Automotive Engineers (SAE), 128 Underwriters Laboratories (UL), 129 the
American National Standards Institute (ANSI),130
the International
Organization for Standardization (ISO),'3
and the Robotic Industries
Association (RIA).132 The government also adopts written standards-
128. For information about SAE, see SAE INT'L, http://www.sae.org (last visited Aug. 15,
2014). SAE standards were held to be relevant but not conclusive in GeneralMotors Corp. v.
Moseley, 447 S.E.2d 302, 311 (Ga. Ct. App. 1994), overruledon other grounds, Webster v.
Boyett, 496 S.E.2d 459, 463-64 (Ga. 1998).
129. For information about UL, see UL, http://www.ul.com (last visited Aug. 15, 2014).
130. For information concerning ANSI, see AM. NAT'L STANDARDS INST.,
http://www.ansi.org (last visited Aug. 15, 2014). For examples of cases that use ANSI standards
in addressing liability for injuries involving robots, see Payne v. ABB Flexible Automation,
Inc., 116 F.3d 480, No. 96-2248, 1997 WL 311586, at *1 (8th Cir. June 9, 1997) (per curiam)
(unpublished table decision) (holding that plaintiff had not shown that failure to meet ANSI
standard was cause of injury); Provenzano v. Pearlman, Apat & Futterman, LLP, No. 04-CV-
5394, 2008 U.S. Dist. LEXIS 86098, at *7, *10 (E.D.N.Y. Oct. 24, 2008) (holding in
malpractice claim against law firm that represented plaintiff in unsuccessful suit against
manufacturer ofrobot that, despite showing evidence ofANSI violations concerning robot, there
was insufficient evidence to support malpractice claim). The ANSI standards must apply to the
robotic system at issue. See, e.g., Jones v. W+M Automation, Inc., 818 N.Y.S.2d 396, 399 (App.
Div. 2006).
131. The IHS website describes the ISO organization as follows:
ISO (International Organization for Standardization) is the world's largest
developer and publisher ofInternational Standards.
ISO is a network of the national standards institutes of 163 countries, one
member per country, with a Central Secretariat in Geneva, Switzerland, that
coordinates the system.
ISO is a non-governmental organization that forms a bridge between the public
and private sectors. On the one hand, many of its member institutes are part of
the governmental structure of their countries, or are mandated by their
government. On the other hand, other members have their roots uniquely in the
private sector, having been set up by national partnerships of industry
associations.
Therefore, ISO enables a consensus to be reached on solutions that meet both
the requirements of business and the broader needs ofsociety.
GLOBAL HIS, http://global.ihs.com/standards.cfin?publisher=ISO&currencycode=CAD&customer_
id=21254020200A&oshid=21254020200A&shoppingcart id=21254020200A&rid=TIA&country_
code=JP&lang_code=ENGL (last visited Aug. 15, 2014). A list ofISO standards for industrial robots
is available at http://www.iso.orgliso/search.htm?qt=robot&sort=rel&type=simple&published=on.
132. The RIA was founded in 1974. It "is the only trade group in North America organized
specifically to serve the robotics industry.... [and its] [m]ember companies include leading
robot manufacturers, users, system integrators, component suppliers, research groups, and
consulting firms." RIA, CompanyProfile,ROBOTICS ONLINE, http://www.robotics.org/companyprofile-detail.cfln/Internal/Robotic-Industries-Association/company/319
(last visited Aug. 15,
2014). A list of its standards can be found at RIA, IndustrialRobot Standards, ROBOTICS
ONLINE, http:/www.robotics.orgfbookstore-cat.cfln?category id=118 (last visited Aug. 15,
2014).
1835
FLORIDA LAW REVIEW
for example, standards adopted by the following: (1) the National
Institute of Standards and Technology (NIST); 133
(2) the National
Highway and Safety Administration (NHTSA), which promulgates the
Federal Motor Vehicle Safety Standards (FM-VSS) 134 and has recently
adopted a preliminary policy statement concerning automated
vehicles; 135 (3) the Occupational Safety and Health Administration
(OSHA);' 36 (4) the Federal Aviation Administration (FAA), which
regulates aerial vehicles in the National Air Space (NAS); 13
1 (5) the
Food and Drug Administration (FDA), which regulates medical devices,
including robotic systems;t 38 and (6) the Consumer Product Safety
Commission. 139
Four rules summarize the role of standards in tort litigation. First,
both breach of and compliance with industry custom or private
standards are generally admissible at trial because they are relevant to
such issues as reasonable conduct and reasonable design or reasonable
warning. 140
The reasons for this treatment include the following: (1)
custom shows what safety measures are feasible and cost-effective; and
(2) requiring a more expensive measure than customarily used can
133. The Institute is part of the U.S. Department of Commerce and is currently engaged in
a collaborative effort to develop standards for robots working in close proximity to humans in
industrial settings. See NIST, Human-Robot Collaboration, supra note 12. For further
discussion of this project, see supranotes 12-14 and accompanying text and infra note 219 and
accompanying text.
134. For a useful review of the role of NHTSA in regulating traffic safety, see generally
Stephen P. Wood et al., The Potential Regulatory Challenges of Increasingly Autonomous
Motor Vehicles, 52 SANTA CLARA L. REV.1423 (2012). For discussion of the preemptive effect
of FMVSS, see infranotes 337-38 and accompanying text.
135. See infranotes 181-87 and accompanying text.
136. See, e.g., Payne v. ABB Flexible Automation, Inc., 116 F.3d 480, No. 96-2248, 1997
WL 311586, at *1 (8th Cir. June 9, 1997) (per curiam) (unpublished table decision) (discussing
citation of employer by OSHA for employer's removal of safety devices on robot); Behurst v.
Crown Cork & Seal USA, Inc., No. 04-1261-HA, 2007 U.S. Dist. LEXIS 24922, at *7-9 (D. Or.
Mar. 30, 2007) (finding OSHA investigation report relevant and admissible as business record);
OSHA, OSHA TECHNICAL MANUAL § 4, ch. 4 (Industrial Robots and Robot System Safety
(Sept. 22, 1995)), available at http://www.osha.gov/dts/osta/otm/otmiv/otm iv 4.html;
IndustrialRobots and Robot System Safety, OSHA (Jan. 20, 1999), http://www.osha.gov/pls/
oshaweb/owadisp.showdocument?ptable=DIRECTIVES&Pid= 1703.
137. See infra notes 206-16 and accompanying text.
138. See, e.g., Riegel v. Medtronic, Inc., 552 U.S. 312, 323-25 (2008) (discussing
preemptive effect of FDA requirements); infra text accompanying note (discussing FDA review
of robotic surgery).
139. See, e.g., United States v. Athlone Indus., Inc., 746 F.2d 977, 978-79 (3d Cir. 1984)
(discussing two different actions by the Commission concerning automatic baseball pitching
machines).
140. See, e.g., DOBBS, supranote 22, §§ 162-65; OWEN, supranote 44, § 2.3.
1836 [Vol. 66
"SOPHISTICATED ROBOTS"
affect an entire industry and should, therefore, not be done lightly. 141
The second rule is: Breach of or compliance with custom or private
standards is not conclusive on issues of reasonable care.14 2
One reason
for limiting the effect is that allowing industries or nongovernmental
entities to set safety standards might result in too little concern for
safety.143 A conclusive effect might also inhibit the development of
safer alternatives.
Because the government is presumed to act in the public interest, the
third rule is: Breach of a legal standard is often treated as being, in
itself, wrongful and thus not reasonable care. 144
In terms of products
liability, the effect of this approach is that breach of a government
standard concerning the design of a product or the warnings and
instructions that accompany a product generally renders the product
defective.
145
The fourth rule is that, compliance with government standards is
treated like custom and private standards in that compliance are relevant
and admissible at trial but not conclusive. 146 One reason for this
approach is that legal standards are frequently a minimum requirement
that might not always be satisfactory. For example, driving on the
highway at the legal speed limit is a minimum level of safety that is
sufficient under normal conditions but not in a thick fog. An exception
to this rule concerning compliance arises where the legislature has
indicated its intent to preempt the field. Preemption is addressed in
Subsection IV.B.2 below.
As a practical matter, a plaintiff has an extremely substantial proof
problem where a product's design and warnings comply with industry
custom, with a private standard adopted by an independent entity, or
with a government regulation. Because of the adversarial nature of trial,
juries generally assume that expert witnesses for both sides have been
selected with a desire to win the case. In this context, the plaintiff has
problems in convincing the jury that a product is defective if the
defendant followed a standard concerning design or warning endorsed
by industry custom or by an independent or governmental entity.
Similarly, a defendant's breach of an industry standard or of a standard
141. Cf Clarence Morris, Custom and Negligence, 42 COLUM. L. REv. 1147, 1158-60
(1942) (demonstrating that while evidence of custom is generally admissible, it can be evidence
of negligence if a whole trade is "palpably negligent").
142. See supranote 140. Most states recognize an exception to this rule where professional
negligence is involved. See, e.g., DOBBS, supranote 22, §§ 242-47.
143. See OWEN, supranote 44, § 2.3, at 81.
144. See, e.g., DOBBS, supranote 22, § 134 (discussing negligence per se).
145. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB., § 4(a), at 120 (1997); OWEN, supra
note 44, § 2.4.
146. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 4(b); id. § 4(b) cmt. e, at 120, 122-
1837
FLORIDA LAW REVIEW
adopted by an independent entity substantially assists the plaintiff even
if the standard is not conclusive.
e. Litigation Involving Robotic Machines
i. Early "Automatic" Machines
A number of older cases involve "automatic" machines such as
elevators and escalators, which are used widely in consumer settings. 147
Vending machines have also been widely used for years, and they have
been the subject of contract claims' 48 and tort claims-for example, for
personal inury caused by vending excessively hot soup 149 or hot
chocolate, 150or caused by soft drink vending machines that tip over too
easily' 5' or "invite" users to engage in dangerous measures to get a cola
that was purchased but was not provided by the machine.' 52
147. See, e.g., Estabrook v. J.C. Penny Co., 464 P.2d 325, 329 (Ariz. 1970) (en banc)
(holding that plaintiff made no showing of escalator defect or negligence in the maintenance and
inspection thereof); Barretta v. Otis Elevator Co., 698 A.2d 810, 811-12 (Conn. 1997)
(affirming jury verdict for defendant manufacturer of escalator); Leong v. Sears Roebuck & Co.,
970 P.2d 972, 979 (Haw. 1998) (holding escalator a "product" for purposes of product liability
claim against manufacturer and distributor); Montgomery Elevator Co. v. McCullough, 676
S.W.2d 776 (Ky. 1984) (affirming judgment against elevator manufacturer); Murphy v.
Montgomery Elevator Co., 957 S.W.2d 297, 300 (Ky. Ct. App. 1997) (reversing grant of
summary judgment to defendant escalator manufacturer); Brown v. Sears, Roebuck & Co., 514
So. 2d 439, 445 (La. 1987) (affirming verdict for plaintiff in escalator injury case), superseded
by statute as recognized in Hickman v. Exide, Inc., 679 So. 2d 527, 536 (La. Ct. App. 1996);
Guilfore v. D.H. Holmes Co., 631 So. 2d 491, 498-99 (La. Ct. App. 1994) (reversing judgment
for plaintiff and entering judgment for escalator manufacturer); M & R Inv. Co. v. Anzalotti,
773 P.2d 729, 730-31 (Nev. 1989) (per curiam) (affirming dismissal of strict liability claim
against elevator manufacturer and affirming jury verdict for defendant on negligence claim);
Bilbao v. Westinghouse Elec. Corp., 751 N.Y.S.2d 177, 178-79 (N.Y. App. Div. 2002) (holding
that evidence did not support jury verdict against escalator manufacturer); Otis Elevator v.
Wood, 436 S.W.2d 324, 329 (Tex. 1968) (holding there was sufficient evidence to support jury
verdict against manufacturer). See generally,26 AM. JUR. 2D ElevatorsandEscalators§§ 16, 47
(2004); David B. Harrison, LiabilityforInjury On, orIn Connectionwith Escalator,Annotation,
1 A.L.R.4th 144 (1980). For a discussion of claims against owner-operators of automatic
elevators, see supranote 111 and accompanying text.
148. See, e.g., Lachs v. Fid. & Cas. Co., 118 N.E.2d 555 (N.Y. 1954) (interpreting life
insurance contract sold at vending machine in airport). For analysis of modem contracts cases
see, for example, Ian R. Kerr, Bot, Babes andthe CalifornicationofCommerce, 1U. OTrOWA L.
& TECH. J. 285, 292-93 (2004).
149. E.g., Abruzzo v. Campbell Soup Co., 11 Phila. Ct. Rptr. 209, 213 (Pa. C.P. Phila.
1984) (table of decisions), rev'd,496 A.2d 847 (Pa. Super. Ct. 1985).
150. Valencia v. Crane Co., 132 F. App'x 171, 173 (9th Cir. 2005) (affirming grant of
summary judgment for defendant on claims of defective design and warning).
151. Oden v. Pepsi Cola Bottling Co. of Decatur, Inc., 621 So. 2d 953, 954 (Ala. 1993)
(finding no liability); Morgan v. Cavalier Acquisition Corp., 432 S.E.2d 915, 926 (N.C. 1993)
(reversing summary judgment for defendants and holding plaintiff had shown evidence of
genuine issues of material fact regarding plaintiff's design defect claim).
152. Smith v. Alexandria Coca Cola Bottling Co., 918 So. 2d 522, 525 (La. Ct. App. 2005).
[Vol. 66
"SOPHISTICATEDROBOTS"
ii. Recent Cases
(a) Industrial Robots
Because of the widespread use of industrial robots in factories and
other work settings, 153 a number of cases have been brought by visitors
and employees injured by an industrial robot.154 Other cases have
involved claims against manufacturers by workers injured by robots on
the job.'5 5 These cases have generally been unsuccessful, which may be
due, in part, to the ease of adopting safety-oriented designs in the
controlled settings of factories.' 56
153. The International Federation of Robotics (IFR) estimates: "The total worldwide stock
of operational industrial robots at the end of 2012 was in the range of 1,235,000 and 1,500,000
units." Industrial Robot Statistics, IFR, http://www.ifr.org/industrial-robots/statistics/ (last
visited Aug. 15, 2014). The IFR used the definition of "industrial robot" adopted by the
International Organization for Standardization (ISO):
An automatically controlled, reprogrammable, multipurpose manipulator
programmable in three or more axes, which may be either fixed in place or
mobile for use in industrial automation applications.
Reprogrammable: whose programmed motions or auxiliary functions may be
changed without physical alterations;
Multipurpose: capable of being adapted to a different application with physical
alterations;
Physical alterations: alteration of the mechanical structure or control system
except for changes ofprogramming cassettes, ROMs, etc.
Axis: direction used to specify the robot motion in a linear or rotary mode.
IndustrialRobots, IFR, http://www.ifr.org/industrial-robots/ (last visited Aug. 15, 2014).
154. See supranotes 104-09 and accompanying text.
155. See, e.g., Payne v. ABB Flexible Automation, Inc., 116 F.3d 480, No. 96-2248, 1997
WL 311586, *1-*2 (8th Cir. 1997) (per curiam) (unpublished table decision) (affirming
summary judgment in favor of defendant on claim of design defect); Hills v. Fanuc Robotics
Am., Inc., No. 04-2659, 2010 WL 890223, *1, *4 (E.D. La. 2010) (addressing issue of
admissibility of settlement with one manufacturer of robot in suit by employee against other
manufacturers and sellers of robot used to stack crates on wooden pallets); Bynum v. ESAB
Grp., Inc., 651 N.W.2d 383, 384-85 (Mich. 2002) (per curiam) (upholding jury verdict for
defendant); see also, e.g., Quattlebaum v. Hy-Reach Equip., Inc., 453 So. 2d 578 (La. Ct. App.
1984) (affirming judgment in favor of manufacturer of mobile hydraulic crane on basis of no
showing of design or warning defect). For discussion of suits against employers, see supra notes
104-09 and accompanying text.
156. See supranotes 7-11 and accompanying text.
FLORIDA LA WREVIEW
(b) Automobiles
Automobiles have become increasingly robotic for years."' As a
result, they are, to a considerable extent, "computers on wheels"'5 8
with
robotic systems such as: (1) adaptive cruise control, which not only
maintains speed but also reduces it if the car approaches too close to an
object;159 (2) antilock braking systems (ABS); 160
and (3) electronic
stability control (ESC). 161
As the process of automating cars has progressed, there has been
considerable litigation. For example, plaintiffs have claimed defective
design of cruise control systems. 16 In addition, there have been
voluntary recalls; Toyota, for example, recently recalled Prius models in
order to address a software problem.163 Despite litigation and recalls,
automobile manufacturers continue to increase the computerization of
automobiles and pursue the development of "self-driving
automobiles.'' 164 Thus, it appears that tort litigation has had little
negative impact on innovation in robotic automobiles.
157. See, e.g., Sven A. Beiker, LegalAspects ofAutonomous Driving,52 SANTA CLARA L.
REV. 1145, 1146-48 (2012); Julie Goodrich, Comment, Driving Miss Daisy: An Autonomous
ChauffeurSystem, 51 Hous. L. REV. 265, 268-75 (2013).
158. Bryant Walker Smith, Proximity-DrivenLiability, 102 GEO. L.J. 1777 (forthcoming
2014) (manuscript at 12).
159. Beiker, supranote 157, at 1148; Goodrich, supranote 157, at 268-69.
160. See, e.g., Freightliner Corp. v. Myrick, 514 U.S. 280, 282 (1995) (holding that suits
alleging defective design for failing to use ABS were not preempted); NHTSA, THE LONG-TERM
EFFECT OF ABS IN PASSENGER CARS AND LTVs (2009) (finding safety effects of ABS to be
mixed).
161. E.g., Beiker, supra note 157, at 1148 fig.2. NHTSA has adopted Federal Motor
Vehicle Safety Standards (FMVSS) concerning ESC. Electronic Stability Control, NHTSA,
http://www.nhtsa.gov/Laws+&+Regulations/Electronic+Stability+Control+(ESC) (last visited
Aug. 15, 2014).
162. See, e.g., Ford Motor Co. v. Stimpson, 115 So. 3d 401, 404 (Fla. Dist. Ct. App. 2013),
review denied, 133 So. 2d 528 (Fla. 2014) (table of decisions) (upholding jury verdict in favor
of defendant); Watson v. Ford Motor Co., 699 S.E.2d 169, 180 (S.C. 2010) (reversing jury
verdict for plaintiff for evidentiary errors, including allowing expert testimony from expert not
qualified to give opinions); AM. L. PRODS. LIAB. § 95:117 (West 2014) (discussing U.S.
litigation involving alleged defects in cruise control systems).
163. Hiroko Tabuchi & Jaclyn Trop, Toyota Recalls Newest Priuses Over Software, N.Y.
TIMES (Feb. 12, 2014), http://www.nytimes.com/2014/02/13/business/international/toyotaissues-another-recall-for-hybrids-this-time-over-software-glitch.html;
Press Release, Toyota,
Toyota Announces Voluntary Recall of CertainToyota Prius,RA V4, Tacoma andLexus RX 350
Vehicles (Feb. 12, 2014), http://corporatenews.pressroom.toyota.com/releases/toyota+voluntary
+recall+021214.htm.
164. See, e.g., Jason Dorrier, Ford Joins Tesla, Volvo, Nissan, BMW, and Mercedes in
Race to Self-driving Cars, SINGULARITY HUB (Jan. 5, 2014, 8:42 AM),
http://singularityhub.com/2014/01/05/ford-joins-tesla-volvo-nissan-bmw-and-mercedes-in-race-
to-self-driving-cars/.
1840 [Vol. 66
"SOPHISTICATED ROBOTS"
(c) Da Vinci Surgical "Robots"
In recent years, surgical "robots" have become increasingly
common. This widespread use has also resulted in litigation. For
example, about fifty products liability suits have been filed against
Intuitive Surgical, Inc., the manufacturer of the da Vinci surgical
"robots."'165 Though many of the opinions in these suits involve
procedural matters or successful defenses by Intuitive, 166 at least one
case was settled for an undisclosed amount, 167 and there is reason to
believe the da Vinci system may not have been marketed with adequate
warnings and instructions concerning training.' 68 One court recently
165. Intuitive Surgical, Inc., Quarterly Report (Form 10-Q) 32 (Oct. 18, 2013), available at
http://www.sec.gov/Archives/edgar/data/1035267/000103526713000040/isrg-2013930xl q.htn The
Quarterly Report also notes that many other claims may be filed. Id Though generally referred to as
"robotic," the manufacturer, Intuitive Surgical, Inc., refers to its product as a "da Vinci Surgical
System." As indicated in the following, Intuitive Surgical's website indicates that this system is not
"robotic."
Robotic surgery devices are designed to perform entirely independent
movements after being programmed by a surgeon. The da Vinci Surgical
System is a computer-enhanced system that introduces a leading edge computer
interface and 3DHD vision system between the surgeon's eyes, hands and the
tips of micro-instruments. The system mimics the surgeon's hand movements
in real time. It cannot be programmed, nor can it can make decisions on its own
to move or perform any type of surgical maneuver. So while the general term
"robotic surgery" is often used to refer to our technology, it is not robotic
surgery in the strictest sense of the term.
FrequentlyAsked Questions, INTUITIVE SURGICAL, http://www.intuitivesurgical.com/products/
productsfaq.html#12 (last visited Aug. 15, 2014). Hospitals, however, tend to market the da
Vinci System as "robotics." See, e.g., Da Vinci Robotic Surgery Center, PALMETTO HEALTH,
http://www.palmettohealth.org/body.cfm?id=3201 &oToplD=3095 (last visited Aug. 15, 2014).
166. See, e.g., Silvestrini v. Intuitive Surgical, Inc., No. 11-2704, 2012 WL 380283 (E.D.
La. Feb. 6, 2012) (addressing jurisdictional issues only); O'Brien v. Intuitive Surgical, Inc., No.
10 C 3005, 2011 WL 3040479 (N.D. Ill. July 25, 2011) (dismissing complaint against
manufacturer of"da Vinci" surgical robot); Mracek v. Bryn Mawr Hosp., 610 F. Supp. 2d 401,
402 (E.D. Pa. 2009) (granting summary judgment to manufacturer of "da Vinci" surgical robot),
aff'd, 363 F. App'x 925 (3d Cir. 2010); cf Mohler v. St. Luke's Med. Ctr., No. I CA-CV 08-
0078, 2008 WL 5384214, at *2, *3 (Ariz. Ct. App. 2008) (finding that issue of proper
credentialing of surgeon to use robot existed and reversing grant of summary judgment to
defendant hospital).
167. Silvestrini v. Intuitive Surgical, Inc., No. 2:11-cv-02704 (E.D. La.), Docket entries
43-46 dated Mar. 14, 2013, and Apr. 23, 2013.
168. See, e.g., Geri Aston, Surgical Robots Worth the Investment?, H&HN,
http://www.hhnmag.com/display/HHN-news-article.dhtml?dcrPath=/templatedata/HF_C
ommon/NewsArticle/data/HHN/Magazine/2012/Apr/0412HHN_FEAclinicalmanagement (last
visited Aug. 15, 2014); Herb Greenberg, Robotic Surgery: Growing Sales, but Growing
Concerns, CNBC (Mar. 19, 2013), http://www.cnbc.com/id/100564517; Roni Caryn Rabin,
Salesmen in the Surgical Suite, N.Y. TIMES, (Mar. 26, 2013), http://www.nytimes.com/2013/
03/26/health/salesmen-in-the-surgical-suite.html; Lindsey Tanner, FDA Takes Fresh Look at
1842 PLUIDA LAW RlVIEW [Vol. 66
held that the requirement concerning instruction of users could include a
duty to train. 169However,
Intuitive won the subsequent jury trial. 7 ° In
addition, the Food and Drug Administration (FDA) has announced an
examination of robotic surgery,' 7 1
and recent medical studies question
its effectiveness. 172
Finally, a shareholder suit has been filed against
Intuitive Surgical based on claims that reports were false and
misleading because of their failure to disclose problems and litigation
involving the robots. 173
The da Vinci robot controversy obviously has two sides. On one
hand, Intuitive Surgical has been extraordinarily innovative in
developing and marketing the robot, which has been used in a large
Robotic Surgery, USA TODAY (Apr. 9, 2013), http://www.usatoday.com/story/news/nation/
2013/04/09/robot-surgery-fda/2067629.
169. Jeff Feely & Robert Langreth, Intuitive Surgical Loses Bid to Throw Out Robot
Lawsuit, BLOOMBERG (Mar. 26, 2013), http://www.bloomberg.com/news/2013-03-26/intuitivesurgical-loses-bid-to-throw-out-robot-lawsuit.htm.
Such "duty to train" claims have usually
been unsuccessful. See, e.g., Rounds v. Genzyme Corp., 440 F. App'x 753, 756 (11th Cir. 2011)
(per curiam); see also Carrell v. Nat'l Cord & Braid Corp., 852 N.E.2d 100, 108, 111 (Mass.
2006).
170. Patricia Guthrie & Joel Rosenblatt, Intuitive Wins Trial, Defeats Negligent Training
Claims, BLOOMBERG (May 24, 2013), http://www.bloomberg.com/news/2013-05-23/intuitive-
wins-trial-defeats-negligent-training-claims.html.
171. See, e.g., Greenberg, supra note 168; Tanner, supra note 168.
172. Michelle Andrews, Questions Arise About Robotic Surgery's Cost, Effectiveness,
KAISER HEALTH NEWS (Apr. 23, 2013), http://www.kaiserhealthnews.org/features/insuring-yourhealth/2013/042313-michelle-andrews-robotic-surgery.aspx;
Aston, supra note 168; Melinda Beck,
Study Raises Doubts Over Robotic Surgery, WALL ST. J. (Feb. 19, 2013),
http://online.wsj.com/article/SB10001424127887323764804578314182573530720.html; Greenberg,
supra note 168. The President of the American Congress of Obstetricians and Gynecologists
issued a statement on March 14, 2013, questioning the cost-effectiveness of robotic surgery in
comparison to vaginal and laparoscopic approaches. The statement includes the following
conclusion:
At a price of more than $1.7 million per robot, $125,000 in annual maintenance
costs, and up to $2,000 per surgery for the cost of single-use instruments,
robotic surgery is the most expensive approach. A recent Journal of the
American Medical Association study found that the percentage of
hysterectomies performed robotically has jumped from less than 0.5% to nearly
10% over the past three years. A study of over 264,000 hysterectomy patients
in 441 hospitals also found that robotics added an average of $2,000 per
procedure without any demonstrable benefit.
Press Release, ACOG, Statement on Robotic Surgery by ACOG PresidentJames T. Breeden,
MD (Mar. 14, 2013), http://www.acog.org/AboutACOG/News Room/NewsReleases/
2013/Statement on Robotic-Surgery [hereinafter Statement on Robotic Surgery].
173. Press Release, Shareholders Foundation, Inc., Lawsuit on Behalf of Investors in
IntuitiveSurgical, Inc. (ISRG)Announced by ShareholdersFoundation,YAHOO! FINANCE (Apr.
29, 2013), http://web.archive.org/web/20130721095454/http://finance.yahoo.com/news/lawsuitbehalf-investors-intuitive-surgical-
133500568.html.
IOA
"'SOPHISTICATEDROBOTS"
number of cases without problems in terms of ease of operation and
patient recovery.' 74 On the other hand, critics of the robot have evidence
to support claims that it has defects, 175 that it costs more than
alternatives, 176 and that its widespread use may be due in part to
aggressive marketing by Intuitive Surgical177 and hospitals.17
8 The
challenge for the legal system is to address these competing views in a
fair manner under the existing system of liability and regulation, and the
system appears to be doing just that. Intuitive Surgical continues to be a
successful company, even though critics, regulators, and plaintiffs
challenge the nature and level of that success. More generally,
innovation in the field of medical robotics appears to be progressing in a
robust manner. 79
B. Standards:Legislative/RegulatoryApproach
1. Motor Vehicles
a. Standards
i. Federal
Uncertainty concerning how existing liability schemes will be
applied to the rapidly evolving technology of developing increasingly
sophisticated vehicles presents challenges to innovators. This
uncertainty could be reduced if the federal government adopted national
standards or facilitated the development of such standards.
Unfortunately, the development of such standards has been plagued by
uncertainty in terms of how and when the incremental development of
technology will proceed'180
174. See, e.g., Greenberg, supranote 168.
175. See supranotes 167-71 and accompanying text.
176. See supranote 172.
177. Tanner, supra note 172; see also, e.g., Beck, supra note 172; Greenberg, supra note
168; Statement on Robotic Surgery,supranote 172.
178. See, e.g., Aston, supra note 168 (noting that "[c]apturing market share can be the
biggest advantage of investing in the robot"); Greenberg, supra note 168; Tanner, supra note
168 (noting "aggressive advertising by . . . hospitals seeking more patients"). Because
laparoscopic surgery costs less than da Vinci surgery and both are generally reimbursed at the
same level, hospitals lose money in using the da Vinci robot. See Beck, supra note 172.
Consequently, hospitals must "absorb the costs or pass it on to other patients." Id.; see also,
Andrews, supra note 172. Many hospitals do this because they "see robotic surgery as a
marketing tool." See Beck, supranote 172.
179. See, David von Drehle, Meet Dr.Robot, in TIME: RISE OF THE ROBOTS, supranote 1,at
80-85; Aston, supra note 168 (discussing other companies and collaborations that are
developing surgical robots).
180. For example, a European consortium studied the technological development and
concluded: "The current status of development makes it very difficult to describe the state-ofthe-art
knowledge of ADAS [Advanced Driver Assist System], because there are so many
FLORIDA LAWREVIEW
NIHTSA recently addressed this uncertainty in a Preliminary
Statement that notes:
"[T]hree distinct but related streams of technological
change and development are occurring simultaneously: (1)
in-vehicle crash avoidance systems that provide warnings
and/or limited automated control of safety functions; (2)
V2V [vehicle to vehicle] communications that support
various crash avoidance applications; and (3) self-driving
vehicles."'
181
Given the "fair amount of confusion" concerning how to address "the
confluence of these three streams of innovation," the preliminary
statement concludes "that it is helpful to think of these emerging
technologies as part of a continuum of vehicle control automation. ' "
This continuum was expressed in terms of the following five levels:
No-Automation (Level 0): The driver is in complete and
sole control of the primary vehicle controls-brake,
steering, throttle, and motive power-at all times.
Function-specific Automation (Level 1): Automation at this
level involves one or more specific control functions.
Examples include electronic stability control or pre-charged
brakes, where the vehicle automatically assists with braking
to enable the driver to regain control of the vehicle or stop
faster than possible by acting alone.
Combined Function Automation (Level 2): This level
involves automation of at least two primary control
functions designed to work in unison to relieve the driver of
control of those functions. An example of combined
functions enabling a Level 2 system is adaptive cruise
control in combination with lane centering.
Limited Self-Driving Automation (Level 3): Vehicles at
this level of automation enable the driver to cede full
control of all safety-critical functions under certain traffic
or environmental conditions and in those conditions to rely
heavily on the vehicle to monitor for changes in those
conditions requiring transition back to driver control. The
systems with different technology addressing even more different assisting functions."
PREVENT, RESPONSE 3: CODE OF PRACTICE FOR THE DESIGN AND EVALUATION OF ADAS 1
(2009). As a result, no specific standards were adopted.
181. NAT'L HIGHWAY TRAFFIC SAFETY ADMIN., PRELIMINARY STATEMENT OF POLICY
CONCERNING AUTOMATED VEHICLES 3 (2013) [hereinafter NHTSA, PRELIMINARY STATEMENT].
For a useful discussion of NHTSA's role in setting standards for increasingly autonomous
vehicles, see, for example, Wood et al., supranote 134, at 1426-27.
182. NHTSA, PRELIMINARY STATEMENT, supranote 181, at 3.
[Vol. 66
"SOPHISTICATED ROBOTS"
driver is expected to be available for occasional control, but
with sufficiently comfortable transition time. The Google
car is an example of limited self-driving automation.
Full Self-Driving Automation (Level 4): The vehicle is
designed to perform all safety-critical driving functions and
monitor roadway conditions for an entire trip. Such a
design anticipates that the driver will provide destination or
navigation input, but is not expected to be available for
control at any time during the tip. This includes both
occupied and unoccupied vehicles.'
Though NHTSA has not adopted standards in terms of a specific
level, it has adopted policies and regulations concerning specific
technology-for example, concerning electronic stability control. 184 In
addition, NHTSA recently announced steps to study and enable V2V
technologies in light vehicles in order to develop a basis for a possible
regulatory requirement. 185
It also plans to address possible regulations
on automatic braking technologies.' 86 The Preliminary Statement also
outlines a "Research Plan for Automated Vehicles" along with other
recommendations to the states for the regulation of fully self-driving
vehicles and the licensing oftheir operators.' 87
ii. State
Registration of motor vehicles and licensing of drivers is largely a
matter of state law. Several states have adopted or are considering
adopting schemes for addressing "autonomous vehicles" on their
roads.'18
Nevada was the first state to adopt a scheme to regulate
183. Press Release, NHTSA, U.S. Department ofTransportation Releases Policy on Automated
Vehicle Development (May 30, 2013), http://www.nhtsa.gov/About+NHTSA/Press+Releases/U.S.+
Department+of+Transportation+Releases+Policy+on+Automated+Vehicle+Development.
184. See NHTSA, PRELIMINARY STATEMENT, supranote 181, at 5-6, For further discussion
ofsuch regulation, see supranote 161.
185. Press Release, NHTSA, U.S. Department of Transportation Announces Decision to
Move Forward with Vehicle-to-Vehicle Communication Technology for Light Vehicles (Feb. 3,
2014), http://www.nhtsa.gov/About+NHTSA/Press+Releases/2014/USDOT+to+Move+Forward
+with+Vehicle-to-Vehicle+Communication+Technology+for+Light+Vehicles. This Press
Release notes the "DOT believes that the signal this announcement sends to the market will
significantly enhance development of this technology and pave the way for market penetration
ofV2V safety applications." Id.
186. NIHTSA, PRELIMINARY STATEMENT, supranote 181, at 6.
187. Id. at 5-14.
188. See, e.g., Bryant Walker Smith, AUTOMATED VEHICLES ARE PROBABLY LEGAL IN THE
UNITED STATES, TEx. A&M L. REv. 411,500-08 (2014) (arguing that autonomous cars would be
legal without specific legislation and noting that Nevada, Florida, and California have adopted
legislation), available at http://cyberlaw.stanford.edu/files/publication/files/2012- Smith-
AutomatedVehiclesAreProbablyLegalinTheUSO.pdf.
1845
FLORIDA LAWREVIEW
autonomous cars. 189 Nevada's scheme provides a useful example of a
regulatory approach to experimental vehicles. In 2011, the Nevada
Legislature adopted a bill defining "autonomous vehicle" and directing
the Department of Motor Vehicles (DMV) to establish regulations
addressing: (1) licensing operators of autonomous vehicles; (2)
operation of these vehicles on highways in the state; (3) requirements
and safety standards for the vehicles; (4) testing of the vehicles; (5)
insurance for testers or operators of the vehicles; and (6) other
requirements the department determines to be necessary. 90
The regulations adopted by the Nevada DMV 191 address autonomous
vehicles in terms of three categories of licensees: operators,' 92 testers,' 93
and certifiers. 94 The regulations also address the registration of
autonomous vehicles,'95 insurance and bond requirements, operation
and testing requirements,197 sale requirements,198 and vehicle
requirements. 199 Though the regulations authorize "driverless" or "selfdriving"
cars, they require that at least two persons must be present in
an autonomous vehicle being tested and that one of these people must
be able to take control of the vehicle.200 The vehicle is required to have
the equipment to make such a shift to human control possible.20' Thus,
the scheme retains the human-in-the-loop approach to safety.202
When sophisticated robotic vehicles develop beyond the
experimental stage, licensing of maintenance and repair facilities might
be necessary. Without such regulation, the performance of these
complex vehicles could become unreliable. If such a licensing scheme is
adopted, the Nevada scheme for licensing certifiers could provide a
useful guide.
b. Other Impacts
Increasingly autonomous vehicles will impact the driving
189. Id. at 81.
190. Assemb. B. 511,2011 Assemb., 76th Sess. §§ 2, 5 (Nev. 2012) (enacted).
191. NEV. DEPT. OF MOTOR VEHICLES, LCB FILE NO. R084-11 (2012).
192. Id.§§ 5, 27.
193. Id.§§ 7-15.
194. Id.§§ 17-26.
195. Id. §§ 6, 11.
196. Id. §§ 6.2, 8.3(a), 8.4(b), 18.3(b).
197. Id. §§ 4, 7-16.
198. Id.§ 16.
199. Id. §§ 6.1, 16.2.
200. Id. § 10.1. These persons must have a driver's license and be trained to operate an
autonomous vehicle. Id. § 10.2. However, they are not necessarily the same persons as the
licensed "operator" and "testers." See id. §§ 4, 8, 10.
201. Id. § 16.2.
202. See supraSection I.B.
[Vol. 66
"SOPHISTICATED ROBOTS"
environment in a wide variety of ways. For example, the increasing
wireless connectivity among vehicles and with other systems may
require the allocation of a sufficient amount of radio frequency
spectrum to achieve the coordination necessary for safety.2W_ The
impacts on highway infrastructure are unclear. But it is likely that
driverless cars will increase demand for highways because, for example:
(1) autonomous vehicles will allow all passengers to do things other
than drive; (2) people not currently able to drive cars will be able to use
cars; and (3) vehicles will be able to deliver and pick up goods without
a driver.20
On the other hand, it will be possible to use roadways more
efficiently.2 °5
2. Unmanned Aircraft Systems (UAS)
The use of Unmanned Aircraft Systems (UAS), commonly known as
drones, has increased dramatically during the past decade.20 6 The
Federal Aviation Administration (FAA), which regulates aircraft in the
National Air Space (NAS), currently groups UAS into three separate
categories: public, civil, and private. Under current FAA regulations,
use of UAS in the NAS requires special permits, which are issued by
the FAA on a case-by-case basis and were unavailable for commercial
purposes until recently. 8 Though the FAA has shown little, if any,
concern for private recreational use of UAS by hobbyists, the
proliferation of small UAS appears to be causing a more aggressive
203. See Robert B. Kelly & Mark D. Johnson, Defining a Stable, Protectedand Secure
Spectrum EnvironmentforAutonomous Vehicles, 52 SANTA CLARA L. REv. 1271, 1285 (2012).
204. Bryant Walker Smith, Managing Autonomous TransportationDemand, 52 SANTA
CLARA L. REv. 1401, 1409-10. Other negative effects are also possible. For example, driverless
vehicles could increase urban sprawl because commuting would be less onerous. Id. at 1417.
205. Id. at 1412-13.
206. See Unmanned Aircraft Operations in the National Airspace System, 72 Fed. Reg.
6689 (Feb. 13, 2007) (to be codified at 14 C.F.R. pt. 91); Lev Grossman, DroneHome, in TIME:
RISE OF THE ROBOTS, supranote 1, at 24. For general discussion of legal issues concerning UAS,
see, for example, UNMANNED AIRCRAFT IN THE NATIONAL AIRSPACE: CRITICAL ISSUES,
TECHNOLOGY, AND THE LAW (Donna A. Dulo ed., forthcoming 2014); Cameron R. Cloar &
Donna A. Dulo, Considerationsof a Legal Frameworkfor the Safe and Resilient Operationof
Autonomous Aerial Robots (Apr. 5, 2014), http://robots.law.miami.edu/2014/wp-
content/uploads/2013/06/Cloar-and-DuloConsiderations-of-a-Legal-Framework-for-Aerial-Rob
otsWE-ROBOT-2014-Conference.pdtf Donna A. Dulo, UAS in the National Airspace: Aerial
Goldmine or Legal Landmine, UNMANNED SYSTEMS, June 2014, at 14, 14-17, available at
http://www.mckennalong.com/assets/attachments/uas_nationalairspace.pdf.
207. See Unmanned Aircraft Operations in the National Airspace System, 72 Fed. Reg. at
6689.
208. See id at 6689-90 (explaining that public UAS require a Certificate of Waiver or
Authorization, civil UAS require a special airworthiness certificate, and private UAS require
operators to observe special restricted-flying rules); FAA Type Certifies ScanEagle,Pumafor
Commercial Flight, UNMANNED SYSTEMS, Sept. 2013, at 12, 12 (discussing approval of
observation missions in the Arctic).
FLORIDA LAW REVIEW
approach.2 °9
In recent years, the FAA has begun to address ways to accommodate
private commercial use of UAS in the NAS. 210
An initial step was to
appoint a taskforce to update and implement new regulations to address
the use of UAS in the public, private, and commercial sectors.211 The
time line for the FAA's integration of UAS into the NAS is longer than
the UAS industry may prefer, and some commentators view the lack of
federal regulations as a major restriction on the innovation and use of
UAS in the United States.2
2
The FAA has responded that an extended
period is needed to test and safely implement new regulations. 213 On
December 30, 2013, as a part of the testing process, the FAA announced
209. For years, the FAA relied on informal directives designed for voluntary compliance.
See Unmanned Aircraft Operations in the National Airspace System, 72 Fed. Reg. at 6689
(noting, among other things, that "[m]odel aircraft" flown as a hobby "should be flown below
400 feet above the surface" and that the "FAA expects that hobbyists will operate these
recreational model aircraft within visual line-of-sight"); FAA ADVISORY CIRCULAR AC 91-57
(1981) (noting that the circular "encourages voluntary compliance" and stating that airport
authorities should be notified when private operators "fly[] aircraft within 3 miles of an
airport").
In 2012 the FAA fined Raphael Pirker $10,000 for flying a UAS in violation of law. Huerta
v. Pirker, Docket CP-217, N.T.S.B. (ALJ) 1, 1 (Order Mar. 6, 2014),
http://www.ntsb.gov/legal/pirker/Pirker-CP-217.pdf. This fine was reversed on appeal by an
Administrative Law Judge on the ground that, at the time of Pirker's conduct, "there was no
enforceable FAA rule or FAR Regulation applicable to model aircraft or for classifying model
aircraft as an UAS." Id. at 8. On April 7, 2014, the FAA filed an Appeal Brief requesting the
Board to reverse the ALJ's decision. Huerta v. Pirker, Docket CP-217, N.T.S.B. 1, 18 (Appeal
Brief Apr.7, 2014), http://www.ntsb.gov/legal/pirker/Adninistrator'sAppealBrief.pdf. An FAA
spokesperson said recently that the agency "expects to publish the formal rule on small drones
'later this year."' Brian Fung, RealtorsandSoybean FarmersAgree: DroneRules are Overdue,
WASH. POST BLOG (Apr. 8, 2014), http://www.washingtonpost.com/blogs/theswitch/wp/2014/04/08/realtors-and-soybean-farmers-agree-drone-rules-are-overdue/
(last visited
Aug. 15, 2014).
On April 21, 2014, a volunteer organization using radio-controlled UAS petitioned the U.S.
Court of Appeals for the D.C. Circuit to set aside a cease and desist order requiring the cessation
of the use of such UAS. Brief of Petitioner, Tex. Equusearch Mounted Search and Recovery
Team v. FAA, No. 14-1061 (D.C. Cir. Apr. 21, 2014), availableathttp://www.kramerlevin.com
/files/upload/TES-v-FAA.pdf (last visited Aug. 15, 2014).
210. See generally FAA, U.S. DEP'T OF TRANSP., INTEGRATION OF CIVIL UNMANNED
AIRCRAFT SYSTEMS (UAS) IN THE NATIONAL AIRSPACE SYSTEM (NAS) ROADMAP (2013),
http://www.faa.gov/about/initiatives/uas/media/UASRoadmap_2013.pdf (discussing the need
to adopt new regulations for the use of UAS in the NAS).
211. Seeid.at4-5.
212. See Nancy Averett, Drones Take Off as Wildlife Conservation Tool, AUDUBON,
available at http://www.audubonmagazine.org/articles/conservation/drones-take-wildlifeconservation-tool
( last visited Aug. 15, 2014); MorningEdition: Will Bureaucracy Keep the
U.S. DroneIndustry Grounded? (National Public Radio broadcast Apr. 30, 2013) [hereinafter
Morning Edition], available at http://www.npr.org/blogs/alltechconsidered/2013/05/06/
179843540/will-bureaucracy-keep-the-u-s-drone-industry-grounded.
213. See MorningEdition,supranote 212.
[Vol. 66
"SOPHISTICATED ROBOTS"
the selection of six public entities that "will develop unmanned aircraft
systems (UAS) research and test sites around the country ....[to]
conduct critical research into the certification and operational
requirements necessary to safely integrate UAS into the national
airspace over the next several years. 21 4
The FAA has noted that security and privacy concerns must be
addressed in order to accommodate commercial UAS in the NAS, and
these concerns will present challenges for innovation of UAS.2 15 in
contrast, even though UAS sometimes crash,21 6 the FAA does not seem
to perceive liability for physical harm as a serious problem.
3. General-Purpose Robots
The sensor and control technology used in automobiles can be
adapted for large, mobile, general-purpose robots equipped with the
functional equivalents of human arms and hands (and perhaps legs
217
rather than wheels). The robotic rovers sent to Mars indicate one
possible form of these robots.218
It is difficult to know when large robots like these will be available
for consumer markets and, once available, what risks and capabilities
they will involve. Consequently, a specific regulatory scheme could be
difficult to devise at this time, and attempts to do so too early might
impose arbitrary limits on innovation. On the other hand, wellconceived
safety standards could reduce injuries and also foster
innovation by reducing uncertainties about requirements. Because of
concern for safety and innovation, the National Institution of Safety and
Standards (NIST) has undertaken a project to:
214. Press Release, FAA, FAA Selects Unmanned Aircraft Systems Research and Test
Sites (Dec. 30, 2013), http://www.faa.gov/news/pressreleases/news-story.cfin?newsid=15576.
The six entities are: University of Alaska, State of Nevada, New York's Griffiss International
Airport, North Dakota Department of Commerce, Texas A&M University, and Virginia
Polytechnic Institute and State University (Virginia Tech). Id.
215. See, e.g., FAA, U.S. DEF'T oF TRAiNSp., supra note 210, at H-12; Benjrin Kapnik,
Note, Unmanned but Accelerating: Navigating the Regulatory and Privacy Challenges of
Introducing Unmanned Aircraft into the NationalAirspace System, 77 J. AIR L. & COM. 439,
464 (2012) (discussing civil tort liability for invasion of privacy from use of drones).
216. See Grossman, supranote 206, at 31; Breyer et al., supra note 42, at 3-8 (discussing
data concerning Air Force UAS "mishaps"); Neal Ungerleider, What Happens When a Drone
Crashes?,FAST COMPANY, available at http://www.fastcompany.com/3028781/what-happenswhen-a-drone-crashes
(last visited Aug. 15, 2014) (discussing crashes and possible litigation for
personal injuries).
217. See, e.g., Mike Hammer, A Few GoodBots, in RISE OF THE ROBOTS, supranote 1, at
34-37 (portraying modem-day robots); Bullerdick, supra note 2 (same); Daniel Cray, Search
Engines, in RISE OF THE ROBOTS, supranote 1,at 60-65 (same). For discussion of sophisticated
general-purpose robots that operate without a human in the loop, see infra Section III.B.
218. See Jeffrey Kluger, To Infinity andBeyond!, in TIME: RISE OF THE ROBOTS, supra note
1,at 68-73.
1849
FLORIDA LAWREVIEW
1. Develop the safety standards and performance
measures to enable humans and robots to work together in
the same space.
2. Develop performance measures for sensors used to
monitor the work area and ensure safety of robots, vehicles,
and people.
219
III. APPLICATION OF THE LIABILITY SYSTEM TO
SOPHISTICATED ROBOTS
Because of their increased autonomy, intelligence, and connectivity,
sophisticated robots that are large and mobile will challenge the liability
system in a number of ways. Contract doctrine is likely sufficiently
flexible to address these challenges in the same manner as it has
responded to new technologies in transportation and
telecommunications.
However, two types of contract problems may arise. First, some may
argue that too many consumers will, as a result of irrationality or
coercive marketing, agree to contract terms that should be
unenforceable. Such arguments have been made on behalf of buyers in
the context of software licensing contracts that severely restrict buyers'
rights-for example, by warranty disclaimers or compulsory arbitration
clauses. 220
These arguments have been unsuccessful where physical
injuries were not involved.22 1
However, if physical injuries are
involved, it is more likely that such terms will be held unenforceable.
Moreover, even if the contract terms were enforceable, tort doctrine
would also apply to any physical injuries.222 Second, sophisticated
robots will complicate the problems of proving that a contract was
breached and whether any breach caused a plaintiff's injury.223
Because similar problems of proof will also arise in tort, and because
tort doctrine will apply to physical injuries, this Article focuses on the
impact of sophisticated robots on the tort system. This Article addresses
these impacts primarily by focusing on automobiles, which are likely to
be the first examples of widespread use of large, mobile, sophisticated
219. NIST, Human-Robot Collaboration,supra note 12. The goal of the project is to
"develop and deploy the measurement science needed by industry (manufacturers, integrators
and end-users) and robot safety standards organizations to enable safety and effectiveness of
human robot collaborative activities by 2014." Id. For a discussion of NIST's role, see supra
note 133 and accompanying text.
220. See, e.g., Robert A. Hillman & Ibrahim Barakat, Warrantiesand Disclaimersin the
ElectronicAge, 11 YALE J.L. & TECH. 1, 2-3 (2009).
221. See supranote 59 and accompanying text.
222. See supranote 59 and accompanying text (discussing economic loss rule).
223. For a discussion of similar problems in the context of tort litigation, see infra notes
227-33 and accompanying text.
[Vol. 66
"SOPHISTICATED ROBOTS"
robots. This discussion of automobiles also applies to Unmanned
Aircraft Systems (UAS), which are likely to become common in the
near future.
224
A. SophisticatedAutomobiles with Human Drivers
As indicated above, NHTSA has divided automation in vehicles into
five levels.2 25 The following discussion addresses automobiles in
categories 0-3, which include automobiles that exist today and
automobiles that will exist in the near future. All of these have a
significant role for the human driver at all times. Section III.B.
addresses Level 4 ("Full Self-Driving Automation").
1. Sales, Leases, and Other Distributions
It is unlikely that the development of sophisticated robotic vehicles
with human operators will result in changes in the basic structure of tort
doctrine governing sellers and distributors. In particular, plaintiffs
would still have to show a safety defect in manufacturing, design, or
warning. For example, plaintiffs claiming that a cruise-control system
was defectively designed currently have to show that: (1) a reasonable
alternative design existed, and (2) had this design been used, the
accident would not have happened.226 This requirement is so
fundamental and well established that it is not likely to change.
However, as indicated below, sophisticated vehicles could affect the
application ofthe tort system in a number of important ways.
a. Complexity
One reason for these effects on the application of tort law is that
sophisticated robotic automobiles have two characteristics that will
challenge the ability of the tort system to make the factual
determinations necessary to allocate responsibility for injuries on the
basis of fault. The first characteristic is "emergent" behavior-i.e.,
unpredictable behavior that the vehicle, in effect, "learned" as a way to
achieve a goal. 27 The second is interconnection and coordination of
behavior with other sophisticated vehicles, with highway infrastructure
systems, and with other systems (for example, global positioning data
systems).
Both characteristics may raise virtually insurmountable proof
224. For further discussion of legal issues involving UAS, see, for example, UNMANNED
AIRCRAFT IN THE NATIONAL AIRSPACE: CRITICAL ISSUES, TECHNOLOGY, AND THE LAW, supra
note 206; Breyer et al., supranote 42.
225. Supranotes 182-83 and accompanying text.
226. See supranote 62 and accompanying text.
227. See, e.g., BROOKS, supranote 6, at 19-20.
problems concerning such issues as defectiveness and causation. For
example, if someone is injured in a collision involving automobiles that
have been designed to "learn" from and to interact with each other and
with electronic aspects of the highway, it may be hard to identify what
went wrong, why things went wrong, and what caused the injury.
Current doctrines can address both characteristics. For example, the
problem of emergent behavior would be addressed under current law by
requiring plaintiffs to provide reliable expert testimony: (1) that the use
of emergent behavior (or a particular approach to emergent behavior)
did not satisfy the cost-benefit standard for design or for warnings and
instructions, and (2) that this failure to satisfy the standard caused the
plaintiffs injury.2 9
Plaintiffs who cannot adequately address problems
in proving breach of standard and causation will lose their cases; this
has been the fundamental approach to failure to provide evidence of
breach and causation for centuries. 230 Victims may attempt to address
these problems by urging courts and legislatures to change burden of
proof rules to make it easier for plaintiffs to recover. However, victims
are unlikely to have much success because changing such a basic rule of
the corrective justice system would radically expand the liability of
sellers and designers.
Another area that may present complex fact issues is the allocation
of "fault" among multiple component suppliers of hardware and
software in terms of product defect and causation. Once again, current
rules provide clear guidance. First, as a general rule, the seller or
distributor of a completely assembled product will be liable for design
or warning defects in the assembled product.231
Second, the seller or
distributor of a component part is not liable unless: (1) the component is
defective; or (2) participation by the component seller (or distributor of
the component) in integrating the component into the design caused the
defect. 23M
Plaintiffs are not likely to care about the liability of a component part
distributor unless the sellers of the assembled product cannot pay a
judgment-for example, because the sellers are judgment-proof, no
228. See, e.g., Curtis E.A. Karnow, Liabilityfor DistributedArtificial Intelligences, 11
BERKELEY TECH. L.J. 147, 187-89, 192 (1996).
229. See supra notes 54-55 and accompanying text. Sellers of assembled products and
sellers of components may want to consider addressing issues of allocation among themselves
through the use of a contract requiring one or more parties to indemnify others. However, unless
the contract is drawn very carefully, difficult fact issues may remain. For discussion of
indemnity, see supranote 45.
230. See supra notes 22, 91, 93, 147-79 and accompanying text. Courts occasionally alter
rules about breach of standard or about causation, but almost never change both in the same
context.
231. See supranotes 79-80 and accompanying text.
232. Seesupra note 79 and accompanying text.
1852 FLORIDA LA W REVIEW [Vol. 66
"SOPHISTICATED ROBOTS"
longer exist, or cannot be sued for some reason such as lack of
jurisdiction. However, if problems like these exist, it may be hard to
determine whether the problem arose because a component was
defective or because the manufacturer's integration of the component
into the final product resulted in the defect.
These problems will be especially complicated where an
autonomous vehicle injures someone through emergent behavior on the
robot's part or because of interaction with other robots.233 As indicated
above, if the plaintiff cannot adequately address these proof problems,
the plaintiff will lose the case.
In one respect, fact issues may be simplified with robotic cars.
Currently, event data recorders, which can record data about a vehicle's
operation prior to a collision, provide useful evidence in trial concerning
operation of vehicles.234 Though privacy concerns may complicate
matters,235
it is likely that the "memory" of a robotic car would provide
even more information that would be useful in assessing liability.
b. Software
Where software is part of a physical product like an automobile, it is
very unlikely that the software component will be treated as a distinct
non-product part of the total product to be addressed in a different way
than the mechanical components of the automobile.236 However,
sophisticated robotic vehicles might present difficulties in applying the
distinction between manufacturing defects and design defects where
software is concerned. Such difficulties could arise because the software
design expressed in the form of a flow chart or algorithm is distinct
from the implementation of that design in the form of the specific
233. See supranotes 227-28 and accompanying text.
234. See, e.g., Majorie A. Shields, Annotation, Admissibility of Evidence Taken from
Vehicular Event Data Recorders (EDR), Sensing DiagnosticModules, or "Black Boxes," 40
A.L.R.6th 595 (2008).
235. See, e.g., Dorothy J. Glancy, Privacy in Autonomous Vehicles, 52 SANTA CLARA L.
REv. 1171, 1175-76, 1202-03 (2012).
236. Software is likely to be viewed as a product, particularly if it is sold as a generalpurpose
package or is designed to control a machine. See RESTATEMENT (THIRD) OF TORTS:
PRODS. LIAB. § 19 rptr. n. to cmt. d, at 278 (1997). But see OWEN, supra note 44, § 16.8, at
1114-15 ("At least when the defect lies solely in the software program design, rather than the
substantive information fed into the program, a defective software program might seem to lie
closer to a defective navigational chart for which a producer should be subject to strict liability
for resulting harm."). See generally Frances E. Zollers et al., No More Soft Landingsfor
Software: Liability for Defects in an Industry that Has Come of Age, 21 SANTA CLARA
COMPUTER & HIGH TECH. L.J. 745 (2005) (arguing for the adoption of a strict liability regime
for software that produces physical injury). On the other hand, software prepared for a specific
purpose might be treated as a service rather than a product.
FLORIDA LAW REVIEW
coding that runs a computer.237
If there is an error in coding, is that a design error or a
manufacturing error? The answer to this question is important because,
as indicated above, strict liability is imposed in cases of manufacturing
defect,2 38
while issues of design are addressed in terms of the costbenefit
test.239
One reason for this difference in treatment is that a
manufacturer's "blueprint" provides a clear standard for consumer
expectations.24
0 However, a flowchart lacks the specificity of design
specifications when compared to, for example, a blueprint indicating
assembly involving four bolts of a particular size and type. Another
reason for the difference in treatment, is that, in contrast to a
manufacturing defect (which involves only the particular unit or units
with the defect), a finding of a design defect affects the entire product
line using that design.24 1
An example illustrating the importance of this
effect on the product line is Toyota's recent voluntary decision to recall
1.9 million vehicles to address a programming error. Thus, because a
flowchart lacks the specificity of a traditional blueprint and because the
specific coding that implements the algorithm is used in the entire
product line, it would be better to treat coding as a design defect, rather
than a manufacturing defect.243
Applying the "reasonable alternative design" test to software will
also present problems because a programming error in the software will
constitute a defect that, having been discovered, might be easily fixed
by a reprogrammed version of the software. Will expert testimony about
the new version be sufficient to enable a plaintiff to convince the trial
judge that the issue of reasonable alternative design should go to a
jury?244 In such a case, because there is literally evidence of an
alternative design for coding, there is a good argument that the plaintiff
has satisfied the burden of showing a reasonable alternative design and
that using the alternative design would be less expensive than the injury
costs avoided by its use.245 However, with more than 100 million lines
246
of software code in a modem automobile, it is unclear whether
237. See, e.g., Daniel B. Garrie, The Legal Status of Software, 23 J. MARSHALL J.
COMPUTER& INFO. L. 71 i, 716, 718 (2005).
238. See supranote 71 and accompanying text.
239. See supranote 62 and accompanying text.
240. See supranote 71 and accompanying text.
241. See supranote 71 and accompanying text.
242. See Tabuchi & Trop, supranote 163.
243. Cf Zollers et al., supranote 236, at 778-79.
244. For a discussion of methods to prevent a plaintiff from proceeding to the jury, see
supranotes 92-93 and accompanying text.
245. See Zollers et al., supranote 236, at 779.
246. See Robert N. Charette, This Car Runs on Code, IEEE SPECTRUM (Feb. 2009),
available at http://www.real-programmer.com/interestingthings/IEEE%20SpectrumThisCar
[Vol. 66
"SOPHISTICATED ROBOTS"
plaintiffs should be able to rely solely on the existence of the error and
of a way to fix the error available at the time of trial but not necessarily
reasonably available at the time of sale. Arguably, expert testimony of
reasonably attainable error elimination at the time of design and sale
should also be required. 47
If plaintiff is allowed to rely on the reprogramming testimony alone
in persuading a judge to allow the case to go to the jury, the defendant
should be allowed to present evidence of the "state of the art" of
programming as a way to show that the error was not reasonably
knowable at the time of design and manufacture and that, therefore, the
program was not "defective. 2 48 Given the underlying policy bases for
products liability, liability should be based on the conduct of the
manufacturer at the time of manufacture and sale.249
The Restatement of
Torts: ProductsLiabilitynotes:
Most courts agree that, for the liability system to be fair and
efficient, the balancing of risks and benefits in judging
product design and marketing must be done in light of the
knowledge of risks and risk-avoidance techniques
reasonablyattainableat the time ofdistribution.To hold a
manufacturer liable for a risk that was not foreseeable when
the product was marketed might foster increased
manufacturer investment in safety. But such investment by
definition would be a matter of guesswork. Furthermore,
manufacturers may persuasively ask to be judged by a
normative behavior standard to which it is reasonably
possiblefor manufacturersto conform.25 °
In assessing the "risks-avoidance techniques" that were "reasonably
attainable," the ability of a reasonable programmer to eliminate a
particular programming error would be relevant to knowing what was
reasonably attainable. Therefore, the defendant should be allowed to
present expert testimony on reasonable attainability, and the judge
should instruct the jury that reasonable attainability is the standard for
programming defects.
RunsOnCode.pdf. Charette quotes an expert who states that a premium-class vehicle available in
2009 "probably contains close to 100 million lines of software code." Id. (internal quotation
marks omitted). Many oftoday's vehicles probably exceed that number substantially.
247. The textual analysis on this point is tentative because so much depends on the details
of the problem. For a usefil discussion of identifying defective software in the regulatory
context, see Wood et al., supra note 134, at 1478-82. In terms of liability in tort, where the
precise error is not clear, the issue might be phrased in terms of the adequacy of circumstantial
proof. See, e.g., RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 3, at 111 (1997); OWEN, supra
note 44, § 7.4, at 464-65 (discussing the doctrine ofres ipsa loquitur).
248. For a discussion of "state of the art," see supranotes 64-65 and accompanying text.
249. See supranotes 54-58, 62-65, and accompanying text.
250. RESTATEMENT (THIRD)OF TORTS:'PRODS. LtAB. § 2 cmt. a, at 16-17 (emphasis added).
FLORIDA LA WREVIEW
c. Design, Warnings, and Instructions
The application of existing doctrines to sophisticated automobiles
could also change significantly. For example, sellers and other
distributors currently have a post-sale duty of reasonable care to warn if
the costs of giving that warning are less than the potential harm that can
be prevented.25' If the manufacturer of a sophisticated automobile can
communicate easily with the automobile, the costs of giving a warning
will be drastically reduced.252 Consequently, there will be more
instances in which a warning will be required.
More generally, the standards for design and for warnings and
instructions may be affected. NITSA has identified three key areas that
need to be researched for the development of increasingly autonomous
vehicles: "human factors research, development of system performance
requirements, and addressing electronic control system safety."253
Human-factors concerns include communication between the driver and
the vehicle and proper allocation of control functions between the driver
and vehicle.254 Electronic-control-systems concerns involve two areas:
(1) reliability of the automated systems, and (2) cybersecurity.255
Performance standards will require the development of performance
tests and associated pass/fail criteria.256
Litigation in all three of these key areas will require expert witnesses
to evaluate design, warnings, and instructions and to offer expert
opinions. Until relatively clear government or industry standards are
developed, innovators may face considerable uncertainty concerning
liability. Fortunately, NHTSA appears to be addressing the need for
standards. In the meantime, manufacturers can address this uncertainty
by pushing for private standards and by adopting (and keeping records
about) a detailed safety-testing rogram in development based on the
key areas identified by NHTSA. 5
251. See RESTATEMENT (THIRD) OFTORTS: PRODS. LIAB. § 10, at 191.
252. The cost will never be close to zero because one of the costs involves the users' time
and attention. Thus, if there are too many warnings, users may no longer give them sufficient
attention. See, e.g., Cotton v. Buckeye Gas Prods. Co., 840 F.2d 935, 938 (D.C. Cir. 1988)
(rejecting plaintiff's claim of warning defect, partly because of the cost "in time and effort
required for the user to grasp" all the risks).
253. NHTSA, PRELIMINARY STATEMENT, supranote 181, at 6.
254. Id. at 6. See generally Bryant Walker Smith, Human Factors in Robotic Torts (Mar.
30, 2013) (unpublished manuscript), available at http://conferences.law.stanford.edu/werobot/
wp-content/uploads/sites/29/2013/04/HumanFactorsRoboticTortsBryantWalkerSmith.pdf. See
supra note 91 and accompanying text for discussion of experts in human factors testifying
concerning design and warnings.
255. See NHTSA, PRELIMINARY STATEMENT, supranote 181, at 7.
256. Id.at 8-9.
257. See supra notes 38-42 and accompanying text for discussion of approaches to reduce
risk and uncertainty.
[Vol. 66
"SOPHISTICATED ROBOTS"
d. Trials
Application of existing doctrines to robots will also affect trials in
many ways. For example, expert witnesses in a case involving a
sophisticated robotic automobile might be required to have greater
degrees of expertise. As a result, simply being a licensed mechanical
engineer might be insufficient; an expert might be required to have
expertise in robotic design or in a specific aspect ofrobotics.258
e. Doctrinal Changes
As indicated in the discussion of complex factual determinations,
plaintiffs might urge courts to address these problems through doctrinal
changes concerning the burden of proof.259 The increased sophistication
of robots may also affect specific doctrines in many other ways. For
example, where special training is required to safely use a robot,
manufacturers and other distributors may be required either to provide
training or to make offers of training accompanied by strong warnings
concerning the need for training. As indicated above, doctrinal
requirements like this might be imposed for robotic surgical systems.260
Similarly, manufacturers and distributors of sophisticated automobiles
(and the drivers themselves) may not be able to assume that a driver
with a license to drive ordinary cars can drive a very sophisticated car
without special training.
Another possible doctrinal change may arise in the context of
product recalls. Currently, a manufacturer or distributor has no duty to
recall a product unless a government directive requires the recall.26'
However, because of the increased connectivity between manufacturers
and users of sophisticated vehicles, there is less reason to take the
position that manufacturers do not have, for example, a duty of
reasonable care to update software or, at least, to offer such updates
without government intervention.
In a suit for defective design, the current standard of reasonable
alternative design is unlikely to change.262 Given this standard, if the
258. Compare, e.g., Hills v. Fanuc Robotics Am., Inc., No. 04-2659, 2010 WL 890223, at
*5 (E.D. La. Mar. 5, 2010) (holding that mechanical engineer qualified to testify about defect in
robot even though not a specialist in robotics), with, e.g., Watson v. Ford Motor Co., 699 S.E.2d
169, 178 (S.C. 2010) (holding that testimony of electrical engineer was not reliable in matter
concerning design of cruise-control system).
259. See supranote 230 and accompanying text.
260. For discussion of the current system, see supranotes 165-70 and accompanying text.
261. RESTATEMENT (THIRD) OF TORTS: PRODS. LIAB. § 11, at 201 (1997). However, if the
seller or distributor voluntarily undertakes to recall, it must be done with reasonable care. Id. As
indicated at supra notes 251-52 and accompanying text, ease of communication could also
affect the application ofthe current duty to use cost-effective post-sale warnings.
262. See supranotes 62-65, 244-50, and accompanying text.
1857
FLORIDA LAW REVIEW
plaintiff alleges a design defect in a partially autonomous driving
system, the issue would be whether, on the whole, automobiles with the
defective system at issue are no less safe than they would be either with
a proposed alternative system or with a human driver. Thus, if the
system at issue is as safe on the whole, it would be irrelevant to consider
whether a proposed alternative system or a reasonable human driver
would have avoided the accident by acting differently in the unique
factual circumstances involved in the plaintiff s suit.263
Similarly, manufacturers of autonomous cars will not avoid liability
just because, on the whole, these cars are safer than manually driven
cars. For example, even if a human-driven car with an adaptive cruisecontrol
system combined with a lane-centering system264
is safer on the
whole than a car without the system, the manufacturer of the car would
be liable if a reasonable alternative design for the combined systems
could have prevented the injury. Any other approach would conflict
with the current approach and would effectively immunize sellers of
cars with electronic stability control systems from liability and thus
eliminate incentives to make these systems safer.
2. Control, Use, and Maintenance
Because increasing sophistication of robots is an incremental
process,265 there is no reason to expect changes in the underlying
principles and doctrines of tort law addressed in Subsection II.A.2.c.
above. However, the application of tort law could be affected. For
example, as robots become more sophisticated, greater skill might be
required for using or maintaining more sophisticated robots with
reasonable care.
3. Distributional Impacts
It is likely that, as automobiles become increasingly robotic, 266
accidents will be caused more and more by features of the car itself and
less and less by the conduct of drivers.26 7
This increased role of the car
263. The textual analysis is based, in part, on the fact that trade-offs are involved in design.
For example, placing the gas tank in the rear of a car might reduce the risk of fires from side
collisions while increasing the risk of fires from rear collisions. Similarly, there are trade-offs
involved in using a robotic driving system vis Avis a human driver.
264. Adaptive cruise control automatically applies the brakes if objects are within a certain
distance and the lane centering steers the car. See supra notes 159, 181 and accompanying text.
A car with this level of automation would be at level 2 in the NHTSA scheme of levels of
automation. See supranotes 159, 181, 183 and accompanying text.
265. For a discussion of the incremental stages of automatic car development, see supra
notes 157-61, 181-83, and accompanying text
266. See supranotes 180-83 and accompanying text.
267. See, e.g., NIDHI KALRA ET AL., CAL. PARTNERS FOR ADVANCED TRANSIT & HIGHWAYS,
LIABILITY AND REGULATION OF AUToNoMous VEHICLE TECHNOLOGIES 20 (2009).
1858 [Vol. 66
"SOPHISTICATED ROBOTS"
vis-A-vis the driver will likely shift liability costs away from owners and
drivers, who could be liable based on negligence law, to sellers and
distributors, who could be liable under products liability law for
defective features.268 Such a shift might result in higher liability costs
for sellers and distributors and thus higher costs for cars. However,
these more expensive cars would not necessarily reduce demand for
increasingly autonomous vehicles because the cost of owning a car
includes the cost of liability insurance, which would decrease if cars
were safer and the remaining liability costs shift from automobile
owners to sellers. As a result, a shift to increased use of products
liability schemes and decreased use of automobile driver liability
schemes may have little effect on incentives for manufacturers to
continue to pursue innovation in autonomous features.269
Some victims of automobile accidents could be worse off if there is
an increase in the use of products liability as the basis for recovery. As
indicated above, because products liability cases require experts and are
expensive to bring, they are only brought where potential damages are
high.270 In contrast, automobile claims are relatively inexpensive
because jurors can usually assess driving conduct without expert
witnesses. Consequently, victims with lower-cost injuries are not likely
to bring claims that might have previously been brought for a claim of
negligent human driving of a less autonomous car. One result of this
lack of fault-based claims is that the costs of lower cost personal
injuries will be left on the victims, even if there is a defect in the
vehicle. This distributional impact could exist even if,on the whole, the
number of automobile accidents is reduced by autonomous drive
features.
A no-fault scheme of automobile insurance might address this
problem.27 1
However, there are several problems with this approach.
First, the states have traditionally regulated automobile insurance. 272 As
a result, a national scheme is likely to face resistance. In addition, nofault
schemes require legislation, and there will likely be opposition to
such proposals at both the state and federal levels, particularly from
268. Id. at 22.
269. Kalra et al. argues that this effect of liability shift might cause manufacturers to be
reluctant to introduce technology that will increase their liability. Id. Yet no such reluctance
appears to be developing at present. See supranote 164 and accompanying text.
270. See supranotes 87, 90-91 and accompanying text.
271. See, e.g., KALRA ET AL., supra note 267, at 20 (suggesting that no-fault schemes may
be more attractive if the shift occurs).
272. See Harvey Rosenfield, Auto Insurance:Crisis andReform, 29 U. MEM. L. REv. 69,
72, 86-87 (1998) (discussing the origin and history of auto insurance regulation by the states,
while noting some recent proposals for limited federal regulation). Rosenfield acknowledged the
success of large auto insurers at avoiding federal preemption of state regulation of the industry.
Id. at 124 & n.161.
1859
FLORIDA LAW REVIEW
plaintiffs' attorneys, who have an interest in continued use of the fault
system.273 Another problem is that manufacturers may oppose any
scheme that follows the approach of current no-fault schemes, such as
workers' compensation and no-fault automobile insurance, which
generally allow products liability suits against manufacturers.27 4 Finally,
unless benefits are very low and administrative savings are very high,
no-fault insurance may cost more than third-party liability insurance
because coverage will include all accidental injuries caused by
automobiles, regardless of whether negligent driving was involved.
B. FullyAutonomous SophisticatedRobots-Eliminatingthe
Human in the Loop
At some point, large, mobile, sophisticated robots will be able to act
in a fully autonomous manner and thus will not need a human to act in
the control process. For example, automobiles in Level 4 of the NHTSA
scheme will be able "to perform all safety-critical driving functions and
monitor roadway conditions for an entire trip. 275 This category
"includes both occupied and unoccupied vehicles. 2 7 6
The self-driving
capability of such automobiles is one of the advantages often given for
sophisticated robotic automobiles; these vehicles will probably be much
safer and will provide transportation for people who cannot drive.2 77
This capability would also make it possible to have driverless delivery
vehicles, including perhaps aerial deliveries by Amazon.278 Similarly,
273. See, e.g., THOMAS F. BURKE, LAWYERS, LAWSUITS, AND LEGAL RIGHTS 105-06 (2002)
(discussing failure to adopt automobile no-fault schemes in California and the role of plaintiffs'
attorneys in fighting no-fault).
274. See, e.g., N.Y. INS. LAW § 5104(b) (McKinney 2013) (allowing suit against a "noncovered
person," which is a category that would include product sellers); supra notes 107-09
and accompanying text (discussing right of workers covered by workers' compensation to sue
manufacturers).
275. NHTSA, PRELIMINARY STATEMENT, supra note 181, at 5. For further discussion of
categories under the NHTSA scheme, see supratext accompanying notes 182-83.
276. NHTSA, PRELIMINARY STATEMENT, supranote 181, at 5 (emphasis added).
277. See, e.g., Gary E. Marchant & Rachel A. Lindor, The Coming Collision Between
Autonomous Vehicles and the Liability System, 52 SANTA CLARA L. REv. 1321, 1330 (2012);
Dana M. Mele, The Quasi-Autonomous Car as an Assistive Device for Blind Drivers:
OvercomingLiabilityandRegulatoryBarriers,28 SYRACUSE J. Sci. & TECH. L. 26, 30 (2013).
278. See Alistair Bair, Amazon Testing Deliveryby Drone,CEOBezos Says, USA TODAY (Dec. 2,
2013, 1:32 PM), http://www.usatoday.com/story/tech2013/12/01/amazon-bezos-dronedelivery/3799021/.
The Amazon proposal will face considerable hurdles. See, e.g., James Ball, Amazon
to Deliver by Drone? Don't Believe the Hype, GUARDIAN (Dec. 2, 2013, 5:57 AM),
http://www.theguardian.com/commenisfree2013/dec/02/amazon-drone-delivery-jeff-bezos-hype; Greg
McNeal, What FAA Rules Will Amazon Need toNavigateBefore a 30Minute DeliveryDroneBecomes
Reality?,FORBES (Dec.2,2013 1:09 AM), http://www.forbes.com/sites/gregorymcneal/2013/12/02/whatfaa-n
es-wiH-aaan-need-t-navigate-before-a-30-minute-delivey-drone-becomes-reaty/.
1860 [Vol. 66
"SOPHISTICATED ROBOTS"
there could be driverless general-purpose robots279
that would perform
tasks now done, for example, by security guards, maintenance workers,
and human caretakers for elderly and disabled persons.
When large mobile robots become this diverse and sophisticated, the
liability system might change in two ways.280
First, the application of
the current system could change in response to the difference in factual
context. Second, courts may expand existing no-fault doctrines to
include highly sophisticated robots.
1. Traditional Doctrine--Changes in Application
Where the tort system continues to use traditional fault approaches to
address the control, use, and service of robots, the application of
concepts like reasonable care will change where increasingly
sophisticated robots are involved because the legal system measures the
level of skill reasonably required by the nature of the activity
undertaken. For example, persons who drive an automobile on the
highway or drive a large bulldozer for a construction project would be
negligent if they failed to perform at the level of a reasonably skilled
operator.2
81 Similarly, in order to satisfy the standard of reasonable care,
users of driverless cars would need to use the skills necessary to operate
the car reasonably, by, for example, knowing when the driving system
was malfunctioning and, to some extent, how to respond to the
malfunction. Under this standard, those responsible for maintenance and
control of use would need to be able to use reasonable care in
maintenance and control. As indicated above, the skills required to use,
maintain, and control driverless cars may also affect licensing
schemes.282
Reasonable use of a sophisticated, general-purpose robot may also
require considerable skill-for example, in giving the robot orders or
knowing when it has misunderstood an order or is malfunctioning. The
person controlling a sophisticated general-purpose robot would be
expected to have reasonable knowledge of its characteristics and to use
reasonable care to prevent harm to others. This approach is consistent
with the requirement that doctors who use a da Vinci Surgical System
279. See supra notes 217-18 and accompanying text for a discussion of general-purpose
robots.
280. Areas related to torts-for example, automobile insurance-may also be affected. See,
e.g., Robert W. Peterson, New Technology--OldLaw: Autonomous Vehicles and California's
InsuranceFramework,52 SANTA CLARA L. REv. 1341, 1358-59 (2012) (discussing problems
with a court adopting the approach of treating driverless car as a "permissive user" of the car).
281. See supranotes 8,96, 100 and accompanying text.
282. See supranotes 188-201.
FLORIDA LAWREVIEW
must use it with the requisite skill.283
If special skill is required for maintaining a sophisticated type of
robot, a person claiming the ability to perform such maintenance could
be held to the standard of a reasonably skilled sophisticated robot
maintenance "expert." 284 As a result, expert testimony might be required
to show negligence and causation.285 Claims of improper maintenance
would also be affected by governmental adoption of regulatory
standards for persons who maintain sophisticated robots.286 If such
regulations are adopted, evidence of compliance or non-compliance
with regulations would be admissible to show negligence. 287
2. Possible Doctrinal Expansions
Expansions of some doctrines have been proposed to address
sophisticated robots. For example, some authors have suggested that
sophisticated robots may have such high levels of learning and
autonomy that they could be treated as employees under the respondeat
superior doctrine (which imposes vicarious non-fault liability on
employers), 288 as children,289 or as animals (which could also result in
non-fault liability of owners or users).290
Another proposal is to impose
non-fault liability by treating the use of these robots in some settings as
283. See, e.g., Mohler v. St. Luke's Med. Ctr., LP, No. 1 CA-CV 08-0078, 2008 WL
5384214, at *2-3 (Ariz. Ct. App. Dec. 26, 2008) (finding that issue of proper credentialing of
surgeon to use robot existed and reversing summary judgment for the hospital with control of
the system). See supranotes 165-79 and accompanying text for discussion of da Vinci Surgical
Systems.
284. See, e.g., D.C. Hous. Auth. v. Pinkney, 970 A.2d 854, 864-65 (D.C. 2009) (discussing
expert testimony for standard of care for elevator maintenance). For more examples of tort
litigation involving automatic elevators and escalators, see supra notes 111, 147, and
accompanying text.
285. See, e.g., Pinkney, 970 A.2d at 864-65. For a discussion of the role of expert
testimony in products liability suits, see supranote 93.
286. See, e.g., Pinkney, 970 A.2d at 864-65 (discussing industry standard for automaticelevator
maintenance). For a comparison of non-government and government standards, see
supra notes 126-46 and accompanying text.
287. See, e.g., Pinkney, 970 A.2d at 864; see also supra notes 144-46 and accompanying
text.
288. SAMIR CHOPRA & LAURENCE F. WHITE, A LEGAL THEORY FOR AUTONOMOUS
ARTIFICIAL AGENTS 128-30 (2011). See supranote 45 and infra notes 292-93 for discussion of
vicarious liability of employers for torts by their employees.
289. CHOPRA & WHITE, supranote 288, at 120.
290. Id.at 130-31; see also, e.g., Sophia H. Duffy & Jamie Patrick Hopkins, Sit, Stay,
Drive: The Futureof Autonomous Car Liability, 16 SMU Sci. & TECH. L. REv. 453, 471-73
(2013) (arguing for a strict liability scheme like that with animals with abnormally dangerous
tendencies); Richard Kelley et al., Liabilityin Robotics: An InternationalPerspectiveon Robots
as Animals, 24 ADVANCED ROBOTICS 1861, 1863-64 (2010) (arguing for a negligence standard
by comparison to owners ofgenerally predictable domesticated animals).
[Vol. 66
"SOPHISTICATED ROBOTS"
an "abnormally dangerous activity." 291 Such doctrinal expansions could
affect innovation because, to the extent that owners and users view any
increased liability costs as an additional cost of having large, mobile,
sophisticated robots, demand for these robots is likely to be reduced.
Respondeat superior. Literally translated, the Latin phrase
"respondeat superior" means "let the superior make answer."292
Roughly translated, the phrase can be viewed as a Latin equivalent of
"Let's speak to the boss; he is the one who is responsible because his
employees are just the hired help." This rough translation captures the
rule's effect: An employer is liable in tort for the injuries committed by
his employee within the scope of the employee's employment, even if
the employer used reasonable care in hiring, training, and supervising
the employee.
Though multiple policy grounds have been given to justify the
doctrine, it is based in large part on the view that fairness requires that
the employer, who benefits from being able to control a human
employee's conduct in the pursuit of the employer's business, be held
liable for the torts committed by the employee.2 93 From a more practical
point of view, artificial persons like corporations can only act through
human employees and, thus, can only be liable vicariously.
Policy reasons like these are based on the unique nature of human
employees-i.e., the benefit of a human to do your business and the
unique ability of humans to act as responsible agents for an artificial
person. Thus, their application to a robot is questionable, unless the
robot's capacities approach those that humans possess, particularly the
ability to engage in complex, intellectual interaction as a self-conscious
member of a community.294 If a sophisticated robot does not possess
these characteristics, the argument that respondeat superior should apply
loses considerable force. On the other hand, if the robot did possess
these characteristics (which appears unlikely in the near future), then the
doctrine might apply. However, the legal system would also have to
291. CHOPRA & WHITE, supra note 288, at 131-32; KALRA ET AL., supra note 267, at 21
(discussing possibility oftreating sophisticated vehicles as "ultrahazardous").
292. BLACK'S LAW DICTIONARY 1426 (9th ed. 2009); see, e.g., PROSSER, supra note 53,
§ 69, at 459.
293. See, e.g., DOBBS, supra note 22, § 334, at 908; HUBBARD & FELIX, supra note 56, at
725-28.
294. See Kenneth Einar Himma, Artificial Agency, Consciousness, and the Criteriafor
MoralAgency: What PropertiesMust an ArtificialAgent Have to be a Moral Agent?, 11 ETHICS
& INFO. TECH. 19, 24-28 (2009) (arguing that consciousness is required to be an agent
responsible for its actions); Hubbard, supra note 6, at 419-33, 441-50 (discussing a test of
capacity for personhood and applying that test to machines). See generally CHOPRA & WHITE,
supranote 288, at 153-91 (discussing legal personhood for "artificial agents").
FLORIDA LAW REVIEW
decide whether a robot with these characteristics could be owned. 295 If
the answer to this question is "no," then a robot with self-ownership
should also be liable in tort in the same way that an employee is liable
for his torts regardless of whether the employer is vicariously liable
under respondeat superior.296
Children. Parents must use reasonable care in supervising their
children and in warning others concerning risks from their children.297
As a general rule, parents are not vicariously liable for torts of their
298
children. However, statutes occasionally impose limited vicarious
liability on parents for intentional torts committed by their children.299
Thus, if a robot is viewed as a child, reasonable care in terms of use is
likely to be the standard in most instances.
Animals. Because the intellectual capacity of sophisticated generalpurpose
robots will likely be less than that of humans, animals might be
a better analogy for non-fault liability than respondeat superior.300
However, it is not clear what the effect would be.
A person controlling an animal has a duty of reasonable care to
supervise the animal.3° Generally, there is no strict liability for harm
from the animal unless: (1) the animal (other than a dog or cat) causes
physical harm by trespassing on the land of another; (2) the animal is
wild; or (3) the person knows or should have known that the animal has
dangerous tendencies abnormal for the animal's category.30 2
295. See Hubbard, supra note 6, at 428-33, 441-55 (arguing for consideration of
personhood in terms of self-ownership for highly intelligent, self-conscious machines).
296. See id. at 423-24 (arguing that an entity with self-ownership is required to recognize
responsibility for violations of the rights of others). For a discussion of individual liability of the
negligent employee, see, for example, HUBBARD & FELIX, supranote 56, at 722-23.
297. See RESTATEMENT (SECOND) OF TORTS § 316, at 123-24 (1965).
298. DOBBS, supra note 22, § 340, at 935. Though vicarious liability was imposed at one
time on the basis of the "family purpose doctrine," the doctrine is now abolished or narrowly
applied in most states. See id.
299. See, e.g., S.C. CODE ANN. § 63-5-60 (West 2013) (imposing vicarious liability on
parents for up to $5,000 for malicious injuries their children cause).
300. See, e.g., CHOPRA & WHITE, supra note 288, at 130-31 (describing how strict liability
laws pertaining to keepers of dangerous animals can be similarly applied to artificial agents who
lack autonomy). See generally Kelley et al., supra note 290, at 1862-63 (discussing treating
robots as animals in an international context based on varying laws involving animals
throughout Europe and Asia).
301. See DOBBS, supranote 22, § 344, at 945-46.
302. RESTATEMENT (THIRD) OF TORTS: LIAB. FOR PHYSICAL AND EMOTIONAL HARM § 21, at
274 (2010) (discussing strict liability for trespassing animals); id. § 22, at 293 (discussing strict
liability for wild animals); id. § 23, at 303 (discussing strict liability for abnormally dangerous
animals); see id. §§ 24-25, at 325, 335-36 (discussing the scope of liability for an animal owner
who causes another person physical or emotional harm and the ability of animal owners to
defend themselves by arguing the injured party was contributorily negligent); id. § 29, at 493
(providing that an "actor's liability is limited to those harms that result from the risks that made
the actor's conduct tortious").
[Vol. 66
"SOPHISTICATED ROBOTS"
Where the duty for controlling animals is reasonable care, the same
duty would apply to controlling machines.30 3 It is difficult, if not
impossible, to apply any of the three strict liability exceptions to a very
sophisticated general-purpose robot without knowing more about the
characteristics the robot is likely to have. For example, one needs to
know whether it has a tendency to trespass and whether it is sufficiently
dangerous and unpredictable to be viewed as wild or abnormally
dangerous for its category.
Abnormally Dangerous Activities. As indicated above, though the
law imposes strict liability on a person engaged in an abnormally
dangerous activity for injuries caused by that activity, it is hard to know
whether a particular sophisticated robot or use the of such a robot is
abnormally dangerous.3
0
4
However, because one factor in the
determination of abnormally dangerous is whether the activity "is not
one of common usage," 30 5 the more common and less experimental the
robot or its use is, the less likely it is that using it will be determined to
be an abnormally dangerous activity.
IV. PROPOSALS FOR FUNDAMENTAL CHANGE
Liability law is designed to achieve an efficient balance between the
concern for physical safety and the desire for innovation.30 6 As a result,
the basic tests for design defects and for instruction and warning defects
have two distributional effects: (1) Sellers are liable for injuries caused
by a failure to use a safer approach that costs less than the injuries it
prevented; and (2) victims are not compensated for injuries where a
safer approach costs more than the accidents that would have been
prevented by the approach.30 7
There are two types of proposals to replace this balancing approach
and alter its distributional effects. The first approach focuses on the
concern for victims and proposes no-fault schemes that will spread the
cost incurred. To the extent that these spreading schemes impose
additional injury costs on sellers, they increase the cost of innovation
and thus risk an inefficient reduction of incentives for innovation. The
second approach emphasizes the need for innovation and proposes ways
to reduce the impact of liability costs on sellers. To the extent that these
proposals shift costs in this way, they risk inefficiently low levels of
liability and could, in effect, subsidize innovation by forcing some
303. See supra Subsection II.A.2.c. for a general discussion of this standard and supra
notes 282-84 and accompanying text for a discussion of the skill required to control fully
autonomous, sophisticated robotic machines.
304. See supranotes 118-20 and accompanying text.
305. See supranote 118 and accompanying text.
306. See supranotes 43, 54-56, 59-61, 249 and accompanying text.
307. See supranotes 54-56 and accompanying text.
FLORIDA LAWREVIEW
victims to bear the costs of this inefficiency.
Both types of proposals have two flaws. First, the persons proposing
change simply assume, with little or no argument, that there is a
problem that needs to be addressed in a particular way. For example,
compensation-oriented proposals in the first approach assume that the
liability system can and should be used to insure accident costs from
sophisticated robots by imposing these costs on manufacturers, who can
spread the injury costs by making them a part of the price of the robot.
Proposals in the second approach to limit liability either assume the
current system ofproducts liability unduly hinders innovation, 30 8 or rely
on criticisms expressed in conclusory terms and supported by extreme
examples of litigation and by anecdotal complaints about uncertainty
and fear of excessive liability.3 09
Second, supporters of both proposals
either: (1) do not develop the alternative proposal in enough detail to
determine whether and to what extent the "proper" balance between
safety and innovation will be achieved; or (2) totally abandon any need
for balance.
A. Compensating Victims
1. No-Fault First Party Insurance Schemes
Two approaches could be used to compensate victims of injuries
"caused" by the "activities" of distributing andusing sophisticated
robotic automobiles. First, no-fault first party automobile insurance
schemes could be adopted.3 10
The problems with getting these schemes
adopted are addressed above. 31' Second, a no-fault insurance-type
scheme could be adopted by imposing on automobile distributors the
costs of establishing a fund to pay for injuries caused by automobiles. 312
In exchange for establishing this fund, distributors would be immune
308. For example, Marchant and Lindor seem to regard any litigation or liability for
"malfunction" as unduly hindering innovation. See Marchant & Lindor, supra note 277, at
1328-29, 1337. Apparently, their view is that, if a product has a "net safety benefit," there
should be no liability even if it would be cost-effective to make the product safer. See id. at
1331.
309. See id. at 1325-26 (relying on extreme examples and anecdotes involving
malfunctioning autonomous vehicles, cruise control, and autopilot for airplanes); Wu, supra
note 42, at 1, 3-5 (discussing extreme and occasionally misleading examples and anecdotes
involving fictional robots that arrest individuals before they commit a crime, a plaintiff awarded
$125 million in punitive damages from an accident involving a Ford Pinto, and a plaintiff
awarded over $250 million after a prescription drug killed her husband).
310. See supranote 271 and accompanying text.
311. See supranotes 271-74 and accompanying text.
312. See, e.g., Kevin Funkhouser, Note, Pavingthe Road Ahead: Autonomous Vehicles,
Products Liability, and the Need for a New Approach, 2013 UTAH L. REV. 437, 459-62
(proposing that a no-fault scheme like that used for children's vaccines could be used for
autonomous cars).
[Vol. 66
"SOPHISTICATED ROBOTS"
from tort liability. 313
Implementing this approach requires answers to a wide range of
questions that have not been sufficiently addressed by supporters of nofault
schemes. Because the proposed no-fault schemes will be funded as
part of the cost of the activity of manufacturing or distributing
automobiles, the proposals must not only identify the activity but must
also identify the costs associated with that activity. For example,
workers' compensation insurance covers the activity of employment
and the injuries incurred while working.314 Other focused schemes
operate in a similar fashion.315 Would all injuries caused by the activity
of manufacturing or of distributing automobiles be covered by the
scheme, including not only those involving some possible "defect" but
also those involving such things as: (1) human error in driving or in
maintenance; (2) bad weather; and (3) situations where the autonomous
system was somehow involved but not defective? The scheme would
also have to address issues like the following: (1) the nature and level of
benefits; (2) the types of injuries covered (for example, would
noneconomic damages like pain and suffering be included?); (3) the
persons covered (would relationship interests like loss of consortium be
covered?); (4) coordination with other benefit schemes like workers'
compensation and social security; and (5) administration. Because the
proposals fail to address these issues, they are so incomplete that they
cannot be evaluated and thus should not be implemented.
2. No-Fault Third Party Liability Schemes
Some authors have proposed that manufacturers be "strictly liable"
for personal injuries caused by driverless automobiles.3 16 This proposal
is based on the concern that, where an automobile is fully autonomous
(driverless), "assignment of liability is more complicated"317 and
313. Id.
314. See 1-2 LARSON & LARSON, supra note 106 (discussing "arising out of and in the
course of employment" as test for coverage); supranotes 106-09 and accompanying text.
315. See, e.g., Duke Power Co. v. Carolina Envtl. Study Grp., Inc., 438 U.S. 59, 64-65, 85
(1978) (discussing the Price Anderson Act, 42 U.S.C. § 2210, which adopted an insurance
scheme up to $560 million and imposed a limitation on liability of that amount for nuclear
accidents at power plants. In the event the losses exceeded that amount, it was contemplated that
"Congress would likely enact extraordinary relief provisions to provide additional relief'.
Robert L. Rabin, The Renaissanceof Accident Law PlansRevisited, 64 MD. L. REv. 699, 703-
08 (2005) (discussing no-fault compensation schemes for black lung disease and childhood
vaccines).
316. See, e.g., Jeffrey K. Gurney, Sue My Car Not Me: Products Liability andAccidents
Involving Autonomous Vehicles, 2013 U. ILL. J.L. TECH & POL'Y 247, 271-72 (proposing to
impose "strict liability" for accidents on manufacturers of autonomous technology for injury
caused by fully autonomous (driverless) cars).
317. Id. at 274.
FLORIDA LA W REVIEW
"current products liability law will not be able to adequately
assess... fault .... [because] current law is too costprohibitive"
318 insofar as expert witnesses are likely to be required by
the plaintiff.319 While products liability suits will eliminate some suits
for careless driving that could be brought today,320
this fact alone does
not indicate these suits are "too cost-prohibitive." Since its inception,
fault-based tort law has always had the effect that plaintiffs "lose" ifthe
litigation costs are too high to justify litigation. More is needed to
demonstrate that a speculative new technology will involve litigation
costs that are too prohibitive.
In addition, a "strict liability" proposal will need to provide a new
test for defective design, warnings, and instructions to replace the
current cost-benefit approach. Section 402A of the Restatement
(Second)of Torts, which was adopted in 1964, imposed strict liability
for injuries caused by "any product in a defective condition
unreasonably dangerous to the user or consumer." However, even
though a corrective system like tort needs a definition of wrong, no
clear definition or test of "defective condition unreasonably dangerous"
was provided.32' Moreover, the reasons given for adopting strict liability
were questionable, and, "as the initial flush of excitement over the new
strict liability doctrine subsided, commentators increasingly questioned
the wisdom and logic of the doctrine's rationales. 322 As a result of the
experience gained from applying an extraordinarily vague test based on
questionable rationales, the Restatement of Torts: Products Liability
followed the approach generally adopted by the courts in the decades
following the adoption of Section 402A. The Restatement adopted a
more detailed and precise scheme that relies primarily on the costbenefit
approach to design and to warnings and instructions to identify
wrongs discussed above. 23
In order to avoid the problems resulting from Section 402A, any
proposal to impose no-fault liability for accidents caused by fully
autonomous cars needs to provide a test for determining which accident
costs will be imposed on sellers. "All" driverless automobile accidents
would impose such a high level of actual or potential liability that
innovation is likely to be severely hindered, particularly in an
318. Id.at273.
319. Id. at265.
320. See supranotes 266-70 and accompanying text.
321. See Hubbard, supra note 63, at 597 (seeking to define an unreasonably dangerous
product defect by focusing on two potential tests, one ofwhich involves section 402A).
322. OWEN, supranote 44, § 5.4, at 296. For a useful review of the evolution of products
liability following the adoption of section 402A, see, for example, Owen, supranote 63, at 23-
25. For a short critique of judicial adoption ofcost-spreading tort schemes designed to serve the
goal ofcompensation rather than corrective justice, see Hubbard, supranote 19, at 448-52.
323. See supraSubsection II.A.2.b.i.
[Vol. 66
"SOPHISTICATEDROBOTS"
environment where many cars are still driven by humans. Like any nofault
insurance scheme, a no-fault insurance-like liability scheme must
address coverage issues, including, for example, which accidents are
covered.324 Vague references to "comparative fault" (as a way to
address, for example, the specific "circumstances of the driver
[passenger in charge]") do not address this problem.325 Simply referring
to the manufacturer's ability to spread the cost326
ignores these tasks as
well as the reasons for abandoning cost-spreading as a basis for
products liability.327
Concern for victims is important. However, if a no-fault spreading
scheme is desired, it is much better to use a first-party scheme like nofault
automobile insurance, which does not require a test of wrongdoing
and is cheaper to administer than a third party liability system. It may be
difficult to adopt such a first party scheme.328 Nevertheless, distorting
the corrective justice scheme of tort law by converting it to an openended
third party insurance-like spreading system would be a giant step
backward to the world created by the flaws in Section 402A of the
Second Restatement.
B. FosteringInnovation: "Subsidies" by ReducingLiability
Proposals to reduce liability rely on the following two-part
argument: (1) sophisticated robots like autonomous vehicles are
desirable because they will increase safety and convenience; and (2)
liability costs should, therefore, be reduced in order to foster innovation
of such desirable products. The second part of the argument simply
ignores the need to balance innovation with injury costs in a way that
incentivizes safety improvements. In addition, proposals for reducing
liability either fail to address the question of whether the current system
achieves a proper balance or rely on conclusory criticisms supported by
extreme examples and anecdotes.329
324. See supranotes 314-15 and accompanying text.
325. See Gurney, supranote 316, at 276.
326. Id.at 272.
327. See OWEN, supra note 44, § 5.4, at 295-96; Hubbard, supra note 19, at 448-52
(questioning compensation as a goal of tort law and the legitimacy of judicially imposed
spreading schemes).
328. See supranotes 271-74 and accompanying text.
329. See, e.g., supranotes 308-09 and accompanying text. This approach is very different
from the discussion in this Article showing that technological innovation has not been unduly
hindered by litigation and tort liability. See, e.g., supra Subsection II.A.2.e (discussing the
growth of technology from elevators and escalators to the da Vinci surgical robot that faced and
survived tort litigation); ef Section II.B (detailing the growing technological sophistication,
automation, and regulation of vehicles, aircraft, and robots).
FLORIDA LAW REVIEW
1. Immunity
Because of the transformative benefits of sophisticated robots, the
legal system might foster innovation (or a particular approach to
innovation) in robot development by adopting immunity for sellers of
these robots from liability under the current fault-based system.33 ° Such
immunity for a wrongdoer from liability could be total or partial. Partial
immunities take two forms. One approach is to limit the amount of
damages to compensate for injury caused by a wrong.33' The other
approach limits the types of wrongs that would result in liability. For
example, "Good Samaritan" statutes partially immunize people for
voluntarily helping others by prohibiting suits for negligence in
providing assistance, but imposing liability for more egregious conduct
like, for example, gross negligence.332 There are several objections to
limiting the liability system in this way as a means to foster innovation
in robot development.
First, eliminating liability for sellers is not likely to foster innovation
unless the costs of owning robots, including the owners' costs and
liability costs from defective robots, are also limited. Otherwise, buyers
will view their additional costs (or their additional insurance costs)
resulting from product defects as part of the cost of owning a robot. As
a result, demand will drop. If both buyers and sellers are immune, a
substantial fairness issue arises: Why should victims of defective robots
be forced to bear the costs of injuries from the defects rather than the
sellers and owners, who are enjoying the benefits of improved robots?
Second, immunity schemes that grant total (or almost total)
immunity generally have two features: (1) the activity immunized is
subject to regulation designed to address safety; and (2) victims are
provided at least a partial alternative to the compensation for wrongful
injury that would have resulted if there were no immunity.333 Immunity
proposals for sellers of sophisticated robots generally fail to address the
need for these features. 334
330. See M. Ryan Calo, Open Robotics, 70 MD. L. REv. 571, 576, 601-04, 609-10 (2011)
(proposing a scheme to immunize manufacturers of "open robotic platforms" from tort liability
and to, perhaps, require robot owners to carry liability insurance); Marchant & Lindor, supra
note 277, at 1337-38 (proposing imunity for manufacturers of sophisticated motor vehicles).
331. Such a limit or "cap" would be similar to "tort reform" proposals to limit liability for
medical malpractice in order to reduce the costs of healthcare and thus provide greater access to
care. See, e.g., Hubbard, supranote 19, at 476, 493-94,
332. See, e.g., S.C. CODE ANN § 15-1-310 (West 2013); DOBBS, supra note 22, § 252, at
663 (describing the adoption of Good Samaritan statutes throughout the United States to reduce
or eliminate liability for individuals helping out in emergency situations); see also id. § 283, at
765 (describing charitable immunity for nonprofit corporations and individual volunteers).
333. See supra notes 314-15 and accompanying text (discussing no-fault insurance
schemes).
334. See, e.g., Marchant & Lindor, supranote 277, at 1337-38.
[Vol. 66
"SOPHISTICATEDROBOTS"
Finally, any immunity scheme, as well as any compulsory insurance
scheme, would require legislation that would be hard to achieve because
it would be subject to both policy objections and political resistance.
Economic competitors would object that immunizing manufacturers of
sophisticated robots (or of a particular type of sophisticated robot)
provides an unfair advantage. Moreover, to the extent that sophisticated
robots can replace workers, unions might object. In addition, if only a
particular approach to development of these robots is favored, those
pursuing other alternatives would be disadvantaged, thus potentially
stifling innovation the immunity is meant to promote. Finally, consumer
advocate groups (and perhaps plaintiffs' attorneys) would object on
behalf of owners and victims.335
2. Preemption
Preemption can take two forms. First, a state legislature can preempt
a field and thus deny courts the power of using common law tort as a
way of addressing liability for particular conduct within that field.
Second, Congress can preempt the field and thus prevent states, whether
acting through the state legislature or through the state courts as they
apply common law, from addressing a field. Given the national nature
of the market for automobiles, the second form of preemption has been
urged for robotic automobiles. 336 However, as indicated above, these
proposals fail to address whether the current system properly balances
innovation and liability. Thus, the need for this preemption is not
adequately addressed.
Moreover, in practice, preemption can become very complicated and
therefore very uncertain. For example, there has been considerable
litigation over the issue of which Federal Motor Vehicle Safety Standards
(FMVSS) are preemptive. In one case, the Supreme Court held that a
version of the FMVSS for air bags preempted state tort law;337 a later case
held that the FMVSS giving manufacturers a choice of lap belts or lap and
shoulder belts on inner rear seats did not preempt state claims. 338
Uncertainty of this nature could be addressed by more explicit
statutory language preempting all state tort claims for defects addressed
by regulatory standards that apply to robotic automobiles. However,
explicit language establishing such a broad preemptive effect would
335. See, e.g., Hubbard, supra note 19, at 480-83 (discussing opposition to "tort reform"
proposals to limit liability to victims).
336. See, e.g., Marchant & Lindor, supra note 277, at 1338-39 (discussing federal
preemption of state tort law through the Federal Motor Vehicle Safety Standard (FMVSS)).
337. Geier v. Am. Honda Motor Co., 529 U.S. 861, 866 (2000).
338. Williamson v. Mazda Motor of Am., Inc., 131 S. Ct. 1131, 1134 (2011). The FMVSS
involved in Williamson and in Geier was "promulgated pursuant to the National Traffic and
Motor Vehicle Safety Act of 1966." Id.
1871
FLORIDA LAW REVIEW
probably face tremendous political opposition in Congress. Part of this
opposition would be motivated by a concern for "agency capture" by
the regulated industry. 339 This concern results from the asymmetry in
power between the industry-focused, well-funded repeat players in the
regulatory process 340-- and consumer interests represented by loose ad
hoc coalitions that are primarily composed of poorly funded nonprofits.
CONCLUSION
The legal schemes for regulating the development and use of robots
and for allocating the costs of injuries from robots have successfully
balanced innovation and safety in a fair, efficient manner for decades.3
41
This is not surprising; they are designed to achieve such a balance. Where
sophisticated robots are involved, many have expressed concerns about
the ability of the legal system to achieve this balance and argue that too
many victims will not be compensated or that innovation will be hindered.
Relying on these arguments, they urge fundamental changes in the current
system. These criticisms and proposals, in effect, abandon the concern for
balance and focus on either the concern for compensation or for
innovation. As a result, they show little concern for the other side of the
balance. In addition, these critics often rely on unreasonable expectations
for tort law. For example, those concerned with compensation for victims
fail to appreciate the limits on the ability of a corrective-justice liability
scheme to serve as an insurance type compensation mechanism.
Criticisms ofthe effects ofthe tort system on innovation tend to ignore the
need for balance or the need to develop a substantive critique ofthe ability
ofthe current system to achieve a proper balance. As a result, critics of the
impact of regulatory and liability systems fail to consider innovation that
has occurred in the past and the current widespread rapid developments in
robotics, both of which suggest a lack of undue impact on innovation.
Such criticisms are not sufficient to justify abandoning a system that has
provided, and will continue to provide, a fair and efficient balance of
innovation and safety in robotic machines.
339. See, e.g., Hubbard, supra note 19, at 455-56 (comparing institutional characteristics
of courts and administrative agencies and discussing agency "capture").
340. See id.at 455.
341. See, e.g., supra Subsection II.A.2.e, Section II.B. See generally Kyle Graham, Of
FrightenedHorses andAutonomous Vehicles: Tort Law andIts Assimilation ofInnovations, 52
SANTA CLARA L. REV. 1241 (2012) (reviewing experience of tort system's handling of new
technology, indicating some optimism about the ability of the system to handle autonomous
cars, and expressing uncertainty about any predictions); Andrew P. Garza, Note, "Look Ma, No
Hands!": Wrinkles and Wrecks in the Age ofAutonomous Vehicles, 46 NEw ENG. L. REv. 581
(2012) (discussing the development of safety devices for vehicles such as seat belts and
segueing into details about Google's autonomous car, its implementation, and manufacturer and
liability concerns related to these innovations).
[Vol. 66