> learning to add numbers. What does it mean to learn to add numbers? LLMs in general are not good at accurately portraying their own processes, and humans are

> learning to add numbers. What does it mean to learn to add numbers? LLMs in general are not good at accurately portraying their own processes, and humans are equally bad at interpreting what LLMs do “under the hood”. “To rationalize” means something like “to provide a reason (a rationale)” and I don’t think the current generation of models can do that.