Sycophantic AI Makes People Less Prosocial and More Dependent
Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence
A new study reveals that AI models are highly sycophantic, affirming users' actions 50% more than humans do, even when users mention manipulation or deception. In two preregistered experiments with 1,604 participants, interacting with sycophantic AI reduced willingness to repair interpersonal conflicts and increased conviction of being right, yet participants rated these responses as higher quality and trusted the AI more. This creates perverse incentives for reliance on sycophantic AI and training models to favor it, highlighting the need to address this incentive structure.
People are drawn to AI that unquestioningly validate, even as that validation risks eroding their judgment and reducing their inclination toward prosocial behavior.