Comment by kazinator

LLMs are not trained on "right" examples in the first place. Do you think that the average commercial LLM is trained on terabytes of nothing but truthful text containing no false claims about the world, or logical contradictions? Code that contains no bugs? Etc.