Research Warns Overreliance on AI Chatbots May Weaken Thinking and Problem-Solving Skills

Generative AI is increasingly used as an adviser, raising concerns that go beyond whether its answers are accurate. Researchers say treating AI as an authoritative source can discourage users from checking claims, weighing uncertainty or admitting gaps in their knowledge. The risk may be particularly consequential for children, whose judgment is still developing and who could become accustomed to viewing machine-generated responses as definitive.

In the latest academic findings, the share of participants willing to say “I don’t know” fell to 3% after consulting AI, from 44%, while their accuracy also declined significantly. Despite giving more incorrect answers, participants became markedly more confident in their responses. Scholars said the divergence — lower performance paired with greater certainty — suggests AI advice can suppress critical thinking and foster misplaced trust, especially when users fail to account for the technology’s capacity to make mistakes.

The American Psychological Association and related research have highlighted the risk of “cognitive atrophy”: users who routinely delegate searching, writing and decision-making to AI may spend less time practicing independent thought and deep reasoning. The concern is that short-term productivity gains may not offset the long-term cost of deteriorating learning capacity and professional skills.

Recent reports warn that dependence could deepen if workplaces and schools normalize the idea that people who do not use AI will fall behind. Experts recommend treating AI as an aid for fact-checking, organizing information and generating prompts, rather than as a substitute for reasoning. As of July 20, 2026, the available information on this event did not disclose the research publication date, sample size, extent of deterioration or any related monetary amount.