Using AI even for just 10 minutes can make you mentally lazy, says study

A new study by researchers from MIT, Carnegie Mellon, Oxford and UCLA suggests that using AI chatbots for even 10 to 15 minutes may reduce independent thinking and persistence. While AI improves speed and accuracy in the moment, frequent reliance on direct answers could weaken long-term problem-solving skills.

In Short

• Researchers studied 1,222 participants across three randomised controlled experiments online
• AI-assisted users answered faster initially but struggled more after access ended
• Maths scores fell to 57 per cent versus 73 without AI

Do you use AI? Oh, that would be such an obvious question now, wouldn’t it? So let me rephrase it: how much time do you spend on AI chatbots? Well, if it is close to or more than 10 minutes at a stretch, there is a chance you might slowly be becoming mentally lazy. And no, we are not saying this. The claim comes from researchers at Carnegie Mellon University, MIT, the University of Oxford and UCLA, who suggest that short bursts of AI assistance may reduce a person’s ability to think through problems independently and make them more likely to give up when things become difficult.

In the research titled AI Assistance Reduces Persistence and Hurts Independent Performance, the researchers found that people who used AI tools for around 10 to 15 minutes performed noticeably worse once the AI support was taken away. The study argues that while AI can improve speed and accuracy in the moment, it may also weaken persistence, learning habits and independent problem-solving over time.

The paper was based on a study involving 1,222 participants across three large experiments. Researchers tested participants on maths and reading comprehension tasks using an online platform. Some participants had access to an AI assistant capable of directly solving problems, while others had to complete the tasks on their own.

At first, the AI-assisted group appeared to perform better. They answered questions faster and got more correct answers during the assisted phase. But things changed once the chatbot was suddenly removed. Participants who had relied on AI struggled more when forced to work independently.

In the maths-based experiment, participants who never used AI solved around 73 per cent of problems correctly during the test phase. Those who had earlier depended on AI managed only 57 per cent. A similar pattern appeared in reading comprehension tasks, where the AI-assisted group scored 76 per cent compared with 89 per cent among participants who worked without AI help throughout.

Researchers also noticed another worrying trend: people using AI were far more likely to skip difficult questions or stop trying altogether. The study says the biggest issue was not simply lower scores, but reduced persistence. In simple terms, participants became less willing to struggle through challenges once they got used to instant AI answers.

“AI assistance improves immediate performance, but it comes at a heavy cognitive cost: after just ~10 minutes of AI-assisted problem-solving, people who lost access to the AI performed worse and gave up more frequently than those who never used it,” the paper reads.

In short, the researchers described this as a possible “boiling frog” effect. A single interaction with AI may seem harmless, but repeated dependence could slowly reduce a person’s willingness to think deeply or work through problems independently.

Interestingly, the study found that the negative effects were strongest among people who asked AI for direct answers. Participants who used AI mainly for hints, explanations or clarifications did not show the same level of decline. According to the researchers, this suggests that AI systems designed to coach or guide users may be healthier in the long run than tools that simply complete tasks instantly.

- Ends