See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/366578422
The Use of Python in the field of Artifical Intelligence
Conference Paper · October 2020
CITATIONS
4
READS
937
4 authors, including:
Sinisa Mihajlovic
University of Novi Sad
24 PUBLICATIONS   14 CITATIONS   
SEE PROFILE
Dragan Vojo Ivetic
University of Novi Sad
53 PUBLICATIONS   201 CITATIONS   
SEE PROFILE
Ivana Berković
University of Novi Sad
14 PUBLICATIONS   12 CITATIONS   
SEE PROFILE
All content following this page was uploaded by Sinisa Mihajlovic on 25 December 2022.
The user has requested enhancement of the downloaded file.
The Use of Python in the field of Artifical
Intelligence
S. Mihajlović*, A. Kupusinac**, D. Ivetić**, I. Berković*
* Technical Faculty “Mihajlo Pupin”, Zrenjanin, Republic of Serbia
** Faculty of Technical Sciences, Novi Sad, Republic of Serbia
mihajlovicsinisa25@gmail.com, sasak@uns.ac.rs, ivetic@uns.ac.rs, ivana.berkovic62@gmail.com
Abstract – This paper presents the role of artificial
intelligence and Python programming language. Examples
of Python application in AI fields are described. The
advantages and disadvantages of the Python programming
languages are listed. Because there are a large number of
programming languages used for programming in AI, a
comparison of selected AI programming languages was
performed.
I. INTRODUCTION
College students spend a lot of time learning
programming languages, but they are still not
familiar with all the programming languages that
exist. One of the most popular programing languages
is Python. Python has a wide application, and as such
is often taught in various faculties through subjects.
In the late 1980s, Guido Van Rossum created
Python, which is a powerful, procedural, objectoriented,
and functional programming language. The
language is used in different application domains.
These include software development, web
development, Desktop GUI development, education,
and scientific applications. It thus encompasses all
aspects of development and has gained popularity
thanks to its simplicity and robustness. Python can
be easily learned and understood. Removing the
brackets make the code short. Some tasks in Python
are quite simple, while learning advanced functions
is a bit more complex. The project written in Python
is easy to understand. Python code is concise,
efficient, understandable, and manageable [1].
Artificial intelligence (AI) is widespread field and
is one of the most common subjects in colleges,
where the goal is to introduce students to AI.
John McCarthy is considered the father of
artificial intelligence, and he believes that artificial
intelligence is the science and engineering of making
intelligent machines, especially intelligent computer
programs. AI is a way to make a computer, a
computer-controlled robot, or software that thinks
intelligently, in a similar way that intelligent people
do. By harnessing the power of computer systems, it
has led man to question whether a machine can think
and behave like humans. The development of AI is
intended to create similar intelligence in machines
found in humans [2].
AI is a way to make machines think and behave
intelligently. Machines can be controlled by the
software contained in them, which means that AI has
to do with software programs that control machines.
AI is a science that finds theories and methodologies
that will help machines understand the world and
react to situations in the same ways as humans. AI is
used in many different forms around the world.
Scientists want to make machines feel, judge
thoughts and act, as well as be rational. It is believed
that by understanding how the human brain works,
one can achieve results in the field of AI. A machine
can be built if it imitates the way the human brain
learns, thinks, and acts. All this can be used as a
platform for the development of intelligent systems
capable of learning [3].
Artificial intelligence is considered a trend
technology of the future. There are several
applications that are made on it. It is very interesting
for companies and researchers and raises the
question of which programming language can be
used to develop AI applications. There are various
programming languages such as Lisp, Prolog, C++,
Java, Python, which can be used to develop AI
applications. Among them, the Python programming
language is gaining in popularity, and some of the
reasons are that Python includes very little coding
and simple syntax among other programming
languages that can be developed for AI application
development. Testing for this reason can be easier
and can focus more on programming itself. The main
advantage of using Python for AI is that it comes
with built-in libraries. Python has libraries for almost
all types of AI projects. For example, NumPy, SciPy,
matplotlib, nltk, SimpleAI are some of the important
embedded Python libraries [2].
II. RELATED WORK
Artificial intelligence (AI) as a subfield of
computer science focuses on designing computer
programs and machines capable of performing task
in which people are naturally good, such as
understanding natural language, understanding
speech, and recognizing images [4].
AI strives to create machines are intelligent as
human beings. There are many reasons to study AI.
Nowadays, it works with a huge amount of data,
which the human brain cannot keep track. That is
why it is necessary to automate things. To do
automation, it is necessary to study AI, because it
can learn from data and can perform tasks with
precision, accuracy and without tiredness. It is
necessary for the system to teach itself because the
data itself changes, and the knowledge that results
from such data must constantly updated. That is why
AI is used, because an AI enabled system can teach
itself. With the help of neural networks, AI can
analyze data more deeply, as well as achieve
tremendous accuracy. AI can think and respond to
the situations which are based on the conditions in
real time. For systems that use self-learning
algorithms, the data is intellectual property. AI is
needed to be indexed and organized in a way that
always gives the best results [5].
Intelligence is the ability of a system to calculate,
reason, perceive relationships and analogies, to learn
from experience, store and retrieve information from
memory, solve problems, understand complex ideas,
use natural language fluently, classify, generalize,
and adapt new situations. According to American
development psychologist Howard Gardner,
intelligence comes in many forms: linguistic
intelligence, musical intelligence, logicalmathematical
intelligence, spatial intelligence,
bodily-kinesthetic intelligence, intra-personal
intelligence, interpersonal intelligence. A machine or
artificial intelligence system is considered intelligent
if it possesses at least one of the intelligences.
Intelligence is intangible and consists of reasoning,
learning, problem solving, perception, linguistic
intelligence [2].
AI is a large field of study and helps find
solutions to real-world problems. Fields of study in
AI are machine learning, logic, searching, artificial
neural networks, genetic algorithm, knowledge
representation. AI is applied in gaming, natural
language processing, expert systems, vision systems,
speech recognition, handwriting recognition,
intelligent robots [5].
Python is an example of a language that does
everything right within the domain of things for
which it is designed. What sets Python apart is that
the application can be written on one platform and
can be used on other platforms. Python code is
readable and has a concise syntax that allows
applications to be written using fewer lines of code
than other programming languages. Python supports
functional, imperative, object-oriented, and
procedural coding styles. Python provided for
educational and other purposes for which other
programming languages may fail [6].
Python is particularly attractive for workloads in
data science, machine learning, and scientific
computing. NumPy is the basis of the Python stack
for machine learning. It enables efficient operations
on data structures often used in machine learning,
vectors, matrices, and tensors. The main data
structure of NumPy is a multidimensional array [7].
NumPy is a multidimensional array library with
basic linear algebra routines, while the SciPy library
adorns NumPy arrays with important primitives,
numerical optimizers, and signal processing to
statistics and scarce linear algebra. SciPy is used in
almost half of machine learning projects. The Pandas
library is a format for presenting tabular data in
Python for extracting, transforming, loading context,
and analyzing data [4].
TensorFlow is a software library or framework,
designed by the Google team to apply machine
learning and deep learning concepts in the easiest
way. It combines computational algebra optimization
techniques to easily compute many mathematical
expressions. TensorFlow contains a function that
defines, optimizes, and computes mathematical
expressions using multidimensional arrays called
tensors. Includes deep neural network software and
machine learning techniques. Includes a highly
scalable computing feature with different data sets.
TensorFlow uses GPU computing, automating
management. It includes a unique function of
optimizing the same memory and data used.
TensorFlow is well documented and includes many
libraries for machine learning. It offers several
important functionalities and methods for the same.
TensorFlow is also called a "Google" product. It
includes various machine learning and deep learning
algorithms. TensorFlow can train and run deep
neural networks to manually classify numbers,
recognize images, embed words, and create different
sequence models [8].
Scikit-learn is a Python module that integrates a
wide range of state-of-the-art machine learning
algorithms for supervised and unsupervised mediumsized
problems. It focuses on bringing machine
learning closer to non-specialists who use high-level
general-purpose language. Emphasis is placed on
ease of use, performance, documentation, and
consistency of the API. It has minimal dependencies
and is distributed under a simplified BSD license,
encouraging use in academic and commercial
environments [9].
Theano is a Python library and compilation
optimizer for manipulating and processing
mathematical expressions, which include
multidimensional arrays. Features Theano are tight
integration with NumPy, transparent use of GPU,
efficient symbolic differentiation, optimization of
speed and stability, dynamic C code generation and
extensive unique testing and self-checking [10].
Keras is a deep learning API written in Python,
launched at the top of the TensorFlow machine
learning platform. Allows for quick experimentation.
Keras is an acceptable, highly productive interface
for solving machine learning problems with a focus
on modern deep learning. Provides basic abstractions
and building blocks for the development and
delivery of high-repetition machine learning
solutions. Keras can run on TPU or on large clusters
of GPUs, Keras models can be performed to work in
a browser or on a mobile device [11].
Natural Language Toolkit (NLTK) is a platform
for building a Python program for working with
human language data. It provides easy-to-use
interfaces and lexical resources such as WordNet,
along with a package of word processing libraries for
classification, tokenization, stemming, tagging,
parsing, and semantic reasoning. NLTK is suitable
for linguists, engineers, students, teachers,
researchers, and industry users [12].
III. ADVANTAGES AND DISADVANTAGES OF
PYTHON
Python stood out and became one of the most
popular programming languages. Python includes
very little coding and simple syntax, unlike
programming languages that can be used to develop
AI applications. Testing can be easier, and the focus
can be on programming itself. Python comes with
build-in libraries used for AI. Important
characteristics of Python:
• Support functional and structured
programming methods, as well as OOP.
• It can be used as a scripting language or it
can be compiled into bytecode to build large
applications.
• Provides high-level dynamic data types and
supports dynamic type checking.
• Support automatic garbage collection.
• It can be easily integrated with C, C++,
COM, ActiveX, CORBA, and Java [2].
Like any programming language, python has its
advantages and disadvantages, which are written
below.
A. Advantages of Python
Python is a programming language that has
English-like syntax. It makes it easier to write, read
and understand code. Python is a very productive
language, because due to its simplicity, the focus is
on solving problems. Python is an interpreted
language which means it is executed directly code
by line. In the event of any error, execution is
stopped and the error that occurred is reported.
Python shows only one error, even if there are
multiple errors. This makes it easier to troubleshoot.
Python does not know the type of variable until the
code is run. When executed, it automatically assigns
a data type. The programmer does not have to worry
about declaring variables and their data types.
Python comes under an open source license
approved by OSI, which makes it free to use and
distribute. The source code can be downloaded,
modified, and distributed as own version of Python.
Python’s standard library is huge, and there are
almost all functions for solving various tasks.
Python package manager (pip) facilitates the import
of other packages from the Python package index
(PyPi). Python code can run on different platforms
[13].
B. Disadvantages of Python
Regular code execution often leads to slow
execution. The dynamic nature of Python is
responsible for Python’s low speed because it must
do extra work while executing code. Python
programming language uses a large amount of
memory. It is considered a disadvantage when
creating applications when it is necessary to perform
memory optimization. Python is generally used in
server-side programming. Python is not used on
client-side or mobile applications because it is not
memory efficient and has poor processing power
compared to other languages. Communication with
the database is not Python's strength. The Python
database access layer is primitive and
underdeveloped compared to technologies such as
JDBC and ODBC. Python is a dynamically typed
language, so the data type of a variable can be
changed at any time. A variable containing an
integer can contain a string in the future, which can
lead to execution errors. Therefore, it is necessary to
do a thorough testing of applications [14].
IV. COMPARISON OF SELECTED AI
PROGRAMMING LANGUAGES
There are various languages for writing AI, but
there is no perfect programming language that
would stand out as the best programming language
used in artificial intelligence. The development
process depends on the desired functionality of the
AI application being developed. Based on the
functionality of the AI application, a coding
programming language is used. The languages
commonly used for artificial intelligence projects
are compared to highlight advantages and
disadvantages. Java, Python, Lisp, Prolog and C++
are the main AI programming languages used for
artificial intelligence and that meet different needs
in the development and design of different software.
Python is one of the favorite programming
languages in AI development among developers
because of its syntactic simplicity and versatility.
Python is used for machine learning for developers
because it is less complex compared to C++ and
Java. Python is a portable language because it is
used on platforms including Linux, Windows, Mac
OS, and UNIX. It is likable because of its features
such as interactive, interpreted, modular, dynamic,
portable, and high level which make it more unique
than Java. Python is a multi-paradigm program that
supports object-oriented, procedural, and functional
programming styles. Python supports neural
networks and the development of NLP solutions
thanks to its simple library of functions and more
ideal structure. Supports testing of algorithms
without the need for their application. Python is fast
in development compared to Java and C++.
Developers using Python have difficulty adapting to
a completely different syntax when they need to use
a different language for AI programming. Unlike
C++ and Java, Python works with the help of an
interpreter which makes compilation and execution
slower in AI development. Not suitable for mobile
computing [15].
Python programs run slower than Java programs,
but they also take much less time to develop. Python
programs are usually 3-5 times shorter than
equivalent Java programs. This difference is
attributed to Python's built-in high-level data types
and its dynamic typing. For example, a Python
programmer does not waste time declaring argument
types or variables. The powerful Python
polymorphic list and dictionary type, for which rich
syntactic support is built directly into the language,
find use in almost every Python program. Because
of the run-time typing, Python's run time must work
harder than Java's. For example, when evaluating
the expression, a + b, it must examine objects a and
b to find their type, which is not known at compile
time. The appropriate addition operation is then
called, which can be overloaded with a user-defined
method. Java, on the other hand, can perform integer
addition, but requires declarations of variables for a
and b, and does not allow overloading for the +
operator for instances of user-defined classes. For
these reasons, Python is much better suited as a
"glue" language, while Java is better characterized
as a low-level implementation language [16].
C++ is the fastest computer language. Its speed is
important for developers who value time the most. It
enables fast execution and has less response time
and is therefore applied in search engines and
computer game development. C++ allows extensive
use of algorithms and is efficient in the use of
statistical AI techniques. C++ supports the reuse of
programs in development due to inheritance and
data hiding, which is efficient in saving time and
costs. C++ is suitable for machine learning and
neural networks. It is good for finding solutions to
complex AI problems. It is rich in library functions
and a collection of software tools. C++ is a multiparadigm
program that supports object-oriented
principles that are useful in achieving organized
data. C++ is bad at multitasking. C++ is only
suitable for the application of the kernel or database
of certain systems or algorithms. The bottom-up
approach is extremely complex and makes it
difficult for novice developers to write AI programs
[15]. What applies to Java applies to C++ even
more. The Python code is often 5-10 times shorter
than the equivalent C++ code. Anecdotal evidence
suggests that a single Python programmer can
complete in two months what two C++ developers
cannot complete in a year [16].
Java is a multi-paradigm programming language
that follows object-oriented principles and the Once
Written Read/Run Anywhere (WORA) principle. It
is an AI programming language that can run on any
platform that supports it without the need to
recompile. Most of the syntax is derived from C and
C++. Java is not only appropriate for NLP and
search algorithms but also for neural networks.
Unlike C++, Java is easy to use and even debug. It
has an automatic memory manager that facilitates
the work of programmers. Java is slower than C++,
has lower execution speed and longer response time.
Java is portable, but older platforms would require
software and hardware changes [15].
Lisp is the language used to develop AI. It is the
second oldest programming language after Fortran.
Lisp has evolved into a powerful and dynamic
coding language. Some consider Lisp to be the best
AI programming language because of the liking for
freedom it offers developers. Lisp is used in AI
because of its flexibility for fast prototyping and
experimentation. Lisp has a unique macro system
that facilitates the research and application of
different levels of intellectual intelligence. Lisp is
suitable for inductive logic and machine learning
projects. Lisp supports the compiler instead of the
interpreter and is therefore fast and efficient in
encoding. An automatic memory manager was
invented for Lisp, which collects garbage. Lisp
offers specific control over systems resulting in its
maximum use. Few programmers are familiar with
Lisp programming. Being a vintage programming
language artificial intelligence, Lisp requires
configuration of new software and hardware to
accommodate it use [15]. Lisp is close to Python in
its dynamic semantics, but so different in its
approach to syntax. Python has introspective
possibilities like Lisp. Python programs can
construct and execute program fragments on the fly
[16].
Prolog is also one of the oldest programming
languages and is therefore suitable for AI
development. Like Lisp it is also the primary
language for AI. It has mechanisms that facilitate
flexible frameworks that developers enjoy working
with. It is a declarative language based on rules
because it contains facts and rules that dictate its
language of coding artificial intelligence. Prolog
supports basic mechanisms such as pattern
matching, tree-based data structuring, and automatic
backtracking essential for AI programming. Prolog
has a built-in list processing that is necessary to
represent tree-based data structures. It is effective
for rapidly prototyping frequently published AI
programs. It allows the creation of a database at the
same time as running the program [15].
V. CONCLUSION
Artificial intelligence (AI) is a widespread are
known worldwide. As such, it is often taught
through courses at universities. AI plays a major
role in advancing technology as well as facilitating
various jobs. AI strives to create machines that will
be able to replace people in various jobs, achieving
accuracy and precision. Huge results have been
achieved in the field of AI, but there is still a
tendency towards something bigger and better.
Various programming languages and various
researchers have contributed to this.
Python is used in various fields of AI. Python is
one of the favorite programming languages among
AI developers because of its syntactic simplicity and
versatility. Python is a multi-paradigm program that
supports object-oriented, procedural, and functional
programming styles. There are many programming
languages used for programming in AI. People
constantly wondering which programming language
is best for programming in AI, and there is no
answer to that. Each programming language has its
advantages and disadvantages. This paper presents a
comparison of programming languages that are most
popular among AI programming developers. The list
of programming languages consists of Python, C++,
Java, Lisp and Prolog. Depending on the
functionality of the AI application, a programming
language is used. By selecting the appropriate
language, the appropriate result is achieved.
REFERENCES
[1] H. Bhasin, Python Basics A Self-Teaching Introduction, Mercury
Learning and Information, 2019.
[2] Artifical Intelligence with Python, Tutorials Point (I), Pvt. Ltd.,
2016.
[3] P. Joshi, Artificial Intelligence with Python, Packt Publishing,
January 2017.
[4] S. Raschka, J. Patterson, C. Nolet, Machine Learning in Python:
Main Developments and Technology Trends in Data Science,
Machine Learning, and Artificial Intelligence, Information, April
2020.
[5] W. Ertel, Introduction to Artificial Intelligence, Second Edition,
Springer International Publishing AG, 2017.
[6] J. P. Muller, Beginning Programming with Python for Dummies,
2nd Edition, John Wiley & Sons, Inc., 2018.
[7] C. Albon, Machine Learning with Python Cookbook, O’Reilly
Media, Inc., 2018.
[8] TensorFlow, Tutorials Point (I), Pvt. Ltd., 2018.
[9] F. Pedregosa et al., Scikit-learn: Machine Learning in Python,
Journal of Machine Learning Research 12, pp. 2825-2830, 2011.
[10] Theano. URL: http://deeplearning.net/software/theano/# (accessed
28.09.2019.)
[11] Keras. URL: https://keras.io/about/ (accessed 28.09.2019.)
[12] NLTK. URL: https://www.nltk.org/ (accessed 28.09.2019.)
[13] TechVidvan. URL: https://techvidvan.com/tutorials/pythonadvantages-and-disadvantages/
(accessed 28.09.2019.)
[14] edureka!. URL: https://www.edureka.co/blog/advantages-anddisadvantages-of-python/#Disadvantagesofpython
(accessed
28.09.2019.)
[15] Existek. URL: https://existek.com/blog/ai-programming-and-aiprogramming-languages/
(accessed 28.09.2019.)
[16] Python. URL: https://www.python.org/doc/essays/comparisons/
(accessed 28.09.2019.)
View publication stats