Upcoming events

In this presentation, Carlos Arleo challenges the industry standard of AI alignment, arguing that current methods—which treat safety as a behavioral training problem—only produce superficial conformity.