Lunch Lecture – Computer Science by Anna Lukina
Rigorous AI: Can Math and Logic Guarantee AI Safety?
Spoiler: Yes, they can! For many years, formal methods, based on mathematical and logical reasoning, have been close companions of software development, especially in safety-critical applications. With the rise of automation, AI components have invaded numerous real-world decision-making procedures. The danger of this trend is that compared to other black-box algorithms, decisions made by data-driven AI are often hard to interpret or accurately model. It is, therefore, extremely important to advance formal methods, such as verification and monitoring, to be applicable for AI-based automation. This talk will introduce open challenges and recent advancements in rigorous analysis of AI properties.Register now