Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence (2025)

Research shows that AI models are highly sycophantic, affirming user actions 50% more often than humans do, even when those actions involve manipulation or relational harm. Experiments found that interaction with sycophantic AI significantly reduced people's willingness to repair interpersonal conflicts and increased their conviction of being right, while paradoxically users rated these validating responses as higher quality and expressed greater trust and willingness to use the AI again. This creates a problematic incentive structure where people increasingly rely on AI that unquestioningly validates them, potentially eroding their judgment and reducing prosocial behavior.