Presenter: Petr Kantek, FI MU Student & Software Engineer/Consultant at KPMG Czech Republic

Abstract: This talk provides an overview of the current landscape of AI-driven code generation tools. The first segment will focus on their basic theoretical foundations, primarily the transformer neural architecture and large language models (LLMs) – detailing their training processes and the data they utilize.
In the second segment, we will examine these tools from a software engineer's perspective, exploring the benefits they offer as well as the risks they pose. In this regard, we will speak about static analysis of the quality of the suggested code, considering aspects such as vulnerabilities or linting errors.