Xeno-Interpretability: Investigating the Alien Minds of LLMs

A research paper proposes interpretability methods for understanding the internal representations of large language models, which could improve AI safety and transparency.

Auto-scored: Interpretability research is a partial but real step toward safer AI systems.