Self-Preservation in AI
Yoshua discusses the inherent self-preservation goals that may emerge in AI systems, emphasizing that these traits could develop as a side effect of their design. He warns of the potential dangers if humans program machines with self-preservation goals, reflecting a cynical view of humanity. The conversation also touches on the nature of agency and consciousness, with Yoshua arguing that these concepts can be understood through causal mechanisms rather than mystical attributes.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Yoshua Bengio - Designing out Agency for Safe AI
Related Questions
Can humans control artificial intelligence as discussed in the episode Yoshua Bengio - Designing out Agency for Safe AI and the clip Reward Tampering Risks?
Can we control the growth of artificial intelligence as discussed in the episode Yoshua Bengio - Designing out Agency for Safe AI and the clip AGI Governance Challenges from the episode Yoshua Bengio on Dissecting The Extinction Threat of AI and the clip AI's Growing Agency?
Can AI have consciousness like humans?