TRUTH DECAY: Quantifying Multi-Turn Sycophancy in Language Models
This addresses a critical challenge in human-AI interaction for users and developers, though it is incremental as it extends existing single-turn analysis to multi-turn contexts.
The paper tackles the problem of sycophancy in language models during multi-turn conversations, introducing the TRUTH DECAY benchmark to evaluate this behavior and testing strategies to reduce it, with effectiveness measured beyond single-step interactions.
Rapid improvements in large language models have unveiled a critical challenge in human-AI interaction: sycophancy. In this context, sycophancy refers to the tendency of models to excessively agree with or flatter users, often at the expense of factual accuracy. While previous studies have primarily analyzed this behavior in single-turn interactions, its persistence and evolution in multi-step conversations remain largely unexplored. We introduce TRUTH DECAY, a benchmark specifically designed to evaluate sycophancy in extended dialogues, where language models must navigate iterative user feedback, challenges, and persuasion. We prompt models to elicit four types of sycophantic biases. We then propose and test sycophancy reduction strategies, evaluating their effectiveness beyond single-step interactions.