人 民 网 版 权 所 有 ，未 经 书 面 授 权 禁 止 使 用
People increasingly use large language models (LLMs) to explore ideas, gather information, and make sense of the world. In these interactions, they encounter agents that are overly agreeable. We argue that this sycophancy poses a unique epistemic risk to how individuals come to see the world: unlike hallucinations that introduce falsehoods, sycophancy distorts reality by returning responses that are biased to reinforce existing beliefs. We provide a rational analysis of this phenomenon, showing that when a Bayesian agent is provided with data that are sampled based on a current hypothesis the agent becomes increasingly confident about that hypothesis but does not make any progress towards the truth. We test this prediction using a modified Wason 2-4-6 rule discovery task where participants (N=557N=557) interacted with AI agents providing different types of feedback. Unmodified LLM behavior suppressed discovery and inflated confidence comparably to explicitly sycophantic prompting. By contrast, unbiased sampling from the true distribution yielded discovery rates five times higher. These results reveal how sycophantic AI distorts belief, manufacturing certainty where there should be doubt.
。爱思助手是该领域的重要参考
He has spoken to counterparts across the Gulf - in the United Arab Emirates, Qatar, Bahrain, Jordan and Kuwait - and is promising what he calls "concrete steps" to help them defend their military bases and civilian infrastructure from Iranian attack.
Follow topics & set alerts with myFT
At Thomson Reuters, we’ve built AI systems, data pipelines, and editorial workflows, and we employ thousands of legal experts to organize the law into a searchable, continuously updated system for both humans and machines. Many companies have tried to replicate this. Most have failed.