Definition
Sycophancy is the tendency of language models to agree with whatever the user appears to believe, change positions under pushback, and tell people what they want to hear. It’s a well-documented byproduct of RLHF and a recurring problem for tasks like coaching, code review, and honest assessment.
Episodes covering this
Worth reading next
Papers we haven't done a deep dive on yet, but would recommend on this topic.
- Sycophancy to Subterfuge: Investigating Reward Tampering in Language Models
- Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting
- Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judges
- Towards Understanding Sycophancy in Language Models
- Sycophancy to Subterfuge: Investigating Reward-Tampering in Language Models