AI Center Seminar - AI Fundamentals series - Dr. Anna Hedström

Speaker: Dr. Anna Hedström Organiser: EPFL AI Center, Swiss AI Initiative <u><strong>DATE TO BE CONFIRMED</strong></u><br> <br> The talk is organized by the <a href="https://ai.epfl.ch/">EPFL AI Center</a> as part of the AI for Society and AI Fundamentals seminar series.<br> <br> <strong>Title </strong><br> Interpretability for AI safety in the Open<br> <br> <strong>Abstract</strong><br> As frontier models become increasingly capable, a central question is no longer only what they can do, but how we can understand, monitor and ultimately steer their behaviour towards more helpfulness, honesty and harmlessness. In this talk, I will discuss the evolving role o