How can AIs know what we want if *we* don't even know? (with Geoffrey Irving)
Clearer Thinking with Spencer Greenberg
00:00
Navigating AI Trust and Alignment Challenges
This chapter explores the complexities of developing safe and trustworthy AI systems, highlighting the risks of AI hallucinations and the implications of flawed human training data. It emphasizes the urgent need for effective alignment solutions, balancing AI capabilities with the potential for misuse and manipulation of human emotions. Additionally, the conversation critiques the challenges of understanding and integrating diverse human values in AI training, advocating for a responsible approach to shaping AI behavior.
Transcript
Play full episode
