Hacker News new | ask | show | jobs
by 85392_school 230 days ago
AI is penalized for any and all emitted errors through RL, so this kind of carefulness is actually a strong sign of AI.