AI and Human Conversation скачать в хорошем качестве

AI and Human Conversation

This unmoderated conversation explores artificial intelligence and the potential dangers of AGI, featuring Natasha Vita-More, Eliezer Yudkowsky, Max More, and Anders Sandberg. The event opens with insightful pre-event commentaries, followed by a wide-ranging discussion of key metaphors and frameworks used in AGI risk debates, including black-box and white-box models, the paperclip maximizer, and the role of probability theory. The speakers offer sharply different perspectives: Yudkowsky emphasizes existential risk, Max More argues for more rigorous and nuanced analysis, Anders Sandberg presents a more optimistic vision of human adaptation, and Vita-More introduces alternative, non-absolutist models—such as cooperative inverse reinforcement learning, bounded-range optimization, and bidirectional (reciprocal) alignment—that avoid predetermined outcomes.