Tsvi Benson-Tilsen

Tsvi Benson-Tilsen, a former MIRI researcher, spent seven years grappling with AI alignment challenges. He reveals a stark truth: humanity has made virtually no progress on this complex issue. Tsvi delves into critical concepts like reflective decision theory and corrigibility, illuminating why controlling superintelligence is so daunting. He discusses the implications of self-modifying AIs and the risks of ontological crises, prompting important debates about the limitations of current AI models and the urgent need for effective alignment strategies.