arxiv.org
The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It
Large language models sometimes behave in ways resembling human emotional responses, and recent work has identified internal representations that may explain this. We ask whether LLMs represent pain d...
