Study says relying on AI can turn doubt into false confidence

Researchers led by Valerio Capraro found that deliberately wrong AI answers made people less accurate and more confident on obscure movie questions.

With AI help, participants answered more often, were less likely to say they were unsure, and reported higher confidence than people working alone.

The effect remained even when AI suggestions appeared automatically, and accuracy incentives only partly reduced it.

The study raises concerns that chatbot advice can increase overconfidence and weaken judgment in learning and decision-making.