Dan Williams(@danwilliamsphil):We weren't built by chickens through a process of targeted, intentionally designed reinforcement learning aimed at making us value their interests. If we were - if we evolved to maximise chicken interests instead of Darwinian fitness - it's astronomically unlikely we would treat them in the way that we do now. So it's a very reassuring analogy for thinking about AI risks in the future.

Dan Williams

@danwilliamsphil

Philosopher, University of Sussex. Tweets in personal capacity. Interested in: Philosophy, Psychology, Society. Writes at:

参加 November 2010

3.8K フォロー中 14.2K ファン

Dan Williams@danwilliamsphil

2026.09.10 12:44

We weren't built by chickens through a process of targeted, intentionally designed reinforcement learning aimed at making us value their interests. If we were - if we evolved to maximise chicken interests instead of Darwinian fitness - it's astronomically unlikely we would treat them in the way that we do now. So it's a very reassuring analogy for thinking about AI risks in the future.