Hacker News new | ask | show | jobs
Confessions can keep language models honest (openai.com)
4 points by xavierlint 189 days ago