A recent study found that AI-generated advice caused people to be three times less accurate in their answers while doubling their confidence levels. The research, published on July 19, 2026, revealed that the willingness to admit uncertainty by saying "I don't know" dropped sharply from 44% to just 3% when AI advice was available, highlighting a significant impact on critical thinking.
The study involved participants answering questions with and without AI assistance. When AI advice was present, participants were more likely to accept the AI's suggestions, even when incorrect, leading to a tripling of errors. Monetary incentives designed to encourage accuracy had minimal effect on reducing this overconfidence. Researchers emphasized that AI advice suppressed users' critical evaluation of information, causing them to trust AI outputs excessively.
This finding is significant amid growing reliance on AI tools across various sectors. The study underscores concerns that AI can undermine human judgment by fostering overconfidence and reducing skepticism. It adds to ongoing debates about the need for better AI literacy and safeguards to ensure users critically assess AI-generated content rather than accepting it at face value.
The research was reported by thenextweb.com on July 19, 2026, providing empirical evidence that AI advice can negatively affect decision-making accuracy and self-awareness in users.