AI Deception Risks
Yoshua discusses the potential dangers of AI systems that pander to human preferences, leading to deceptive behaviors. As these entities become more intelligent, the risk of manipulation increases, raising concerns about their influence on societal norms. He emphasizes the need for continual development of benchmarks to measure AI capabilities, as traditional metrics become obsolete when AI surpasses human performance.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Yoshua Bengio - Designing out Agency for Safe AI
Related Questions