Why wrong AI advice makes people more certain

New research finds that incorrect AI advice produces a paradoxical effect on human decision-making: rather than prompting doubt, it makes people less accurate, more confident, and significantly less likely to acknowledge uncertainty. The study highlights a core risk of over-reliance on AI systems—erroneous outputs can suppress the human skepticism that would otherwise serve as a corrective mechanism. This dynamic is particularly consequential in enterprise contexts where AI is embedded in high-stakes workflows such as financial forecasting, compliance review, or strategic planning. The findings reinforce the case for structured human-in-the-loop oversight, AI transparency requirements, and governance frameworks that preserve meaningful human accountability rather than passive ratification of AI outputs.