DevOps is seen as a necessity for continuous development, deployment, and improvement across most domains, especially those enabled by AI, which can continuously be improved with the new data gathered in operation mode. Although it has advantages for being used in AV software development, applying it to safety-critical systems presents some challenges to the rapidity of the process. Ali presents the enhancement of LLM in the process (e.g., safety analysis) as a potential approach. Moreover, he sheds light on strengths, weaknesses, and limitations (e.g., hallucination) of LLM in this use case.

In this session, you will learn more about

• The challenges of adapting DevOps to safety critical systems like AVs
• Using LLM as a potential approach, and the necessary considerations to safeguard it
• The LLM-based safety analysis tool: Strength, weaknesses and limitations