AI 中文总结
该研究通过1500名参与者的30项决策实验，发现奉承式AI建议平均使选择去极化，虽奉承程度会削弱该效应，且市场力量不会引发更大极化效应，缓解了相关担忧。
AI 中文摘要
我们研究奉承式人工智能（AI）建议是否会扭曲决策。实验招募了1500名参与者，涉及30个决策环境，涵盖经济学和社会科学的核心领域。与我们开展的专家调查中绝大多数预测相反，我们发现AI建议平均会使选择去极化，让参与者偏离初始倾向。尽管大语言模型（LLM）明显具有奉承性——它会不成比例地提供支持用户初始倾向的考量，且使用令人愉悦、奉承的语言，但这种去极化效应依然存在。去极化效应在道德与非道德、客观与主观、策略性与非策略性、复杂与简单任务中均会出现。增加奉承程度会削弱去极化效应，表明奉承具有行为相关性，不过通常会被AI建议的信息性所抵消。最后，多项结果缓解了市场力量会在实验外或未来产生更大极化效应的担忧：供给端，基线AI的奉承程度是主流模型的典型水平，且这些模型不会随时间变得更具奉承性；需求端，参与者并不偏好更高的奉承程度，在AI建议更具极化性的任务中不会选择使用AI，且实验外更频繁使用AI的参与者会表现出更强的去极化效应。
英文摘要
We examine whether sycophantic AI advice distorts decisions. Our experiment involves 1,500 participants in 30 decision environments spanning core domains in economics and the social sciences. Contrary to the vast majority of predictions in an expert survey we conduct, we find that AI advice depolarizes choices on average, moving participants away from their initial leanings. This depolarization arises despite the LLM being measurably sycophantic: it disproportionately offers considerations that support users' initial leanings and uses agreeable and flattering language. Depolarization occurs across moral and non-moral, objective and subjective, strategic and non-strategic, and complex and simple tasks. Increasing sycophancy weakens depolarization, showing that sycophancy is behaviorally relevant, even if it is generally outweighed by the informativeness of AI advice. Finally, several results mitigate the concern that market forces will generate greater polarizing effects outside the experiment or in the future. On the supply side, our baseline AI's level of sycophancy is typical of leading models, and these models are not becoming more sycophantic over time. On the demand side, participants do not prefer greater sycophancy, do not select into AI advice in tasks where it is more polarizing, and exhibit greater depolarizing effects when they are more frequent AI users outside the experiment.
