A new study suggests that even brief exposure to AI tools may quickly reduce independent problem-solving performance and persistence.

The study, conducted by researchers at the University of California, Los Angeles, Massachusetts Institute of Technology, the University of Oxford, and Carnegie Mellon University, involved 1,222 participants across three randomized controlled experiments. 

Participants completed math and reading comprehension tasks, with some given access to a chatbot built on GPT-5 while others received no AI assistance.

At first, the AI-assisted group outperformed the control group, answering more questions correctly and completing tasks faster. However, after just 10 to 15 minutes, access to the AI was prohibited.

Once the chatbot was taken away, those who had previously used it showed lower performance. In the math task, the AI-assisted group solved 57 percent of problems, compared with 73 percent among participants who never had AI access. In the reading comprehension task, scores were 76 percent for the AI group versus 89 percent for the control group.

The AI-assisted participants also skipped more questions and showed a greater tendency to give up on difficult problems.

Researchers said the most striking finding was not simply the decline in performance, but the drop in persistence. “It revealed that AI helped in the moment, but when participants returned to working on their own, they performed worse and were more likely to give up,” the analysis noted.

Researchers warned that while AI delivers instant answers and saves time, it may also reduce the need to engage in deeper cognitive effort. 

Skills such as independent reasoning, persistence through complex problems, and confidence built from working through challenges can gradually erode with repeated reliance on AI assistance. They caution that by the time these changes become noticeable, they may already be difficult to reverse.

The researchers described the phenomenon as the “boiling frog” effect, where each instance of relying on AI feels harmless at first, but the accumulated impact on cognitive effort becomes difficult to reverse over time.

“We caution that if such effects accumulate with sustained AI use, current AI systems – optimized only for short-term helpfulness – risk eroding the very human capabilities they are meant to support,” the researchers wrote. 

“People do not merely become worse at tasks, but they also stop trying. If such effects accumulate over months and years of AI use, we may end up creating a generation of learners who have lost the disposition to struggle productively without technological support,” they added. 

Rachit Dubey, a UCLA co-author, warned that rapid AI deployment in education could produce “a generation of learners who will not know what they’re capable of,” potentially diluting human innovation and creativity at scale.

The researchers are also urging AI developers to move away from systems that simply deliver instant, complete answers and instead design tools that guide users through challenges.

“This requires rethinking how AI systems are built to collaborate with humans, and just as the best human collaborators know when not to help, so too should AI,” researchers wrote.