Back to feed
arXiv cs.AI·

A Definition of Good Explanations and the Challenges Explaining LLM Outputs

Signal
65
Hype
15
In three linesPaper proposes a philosophical definition of good explanations based on counterfactual reasoning, accounting for the interlocutor's prior beliefs. Analyzes why LLM outputs are particularly challenging to explain.
Read source
Your take?
ReasoningAI safetyAlignment

Summary generated by Claude — human-verified