A new study suggests that relying on artificial intelligence for simple tasks such as basic maths and reading comprehension could reduce a person’s ability to think independently, even after short periods of use.
The research, released as a preprint, involved 1,200 participants who were asked to complete a series of fraction-based maths problems and reading comprehension exercises. Some participants were allowed to use AI tools for most of the tasks, while others completed the exercises without assistance.
Initial results showed that those using AI performed better when the technology was available, solving more questions correctly. However, when access to AI was removed, their performance declined noticeably. Participants who had relied on AI were more likely to answer questions incorrectly or skip them altogether compared to those who had worked through the problems independently from the start.
Researchers said the findings point to a potential dependency effect, where short-term gains in accuracy come at the cost of reduced cognitive engagement. Even brief reliance on AI appeared to weaken problem-solving skills, raising concerns about how widespread use of such tools could affect learning and intellectual development over time.
The study’s authors compared the effect to a gradual process in which small, seemingly harmless changes accumulate into more significant consequences. They warned that frequent dependence on AI for routine thinking tasks could lead to long-term challenges, particularly in education and skill development.
The results also highlight a broader debate about the role of AI in everyday life. While the technology can improve efficiency and provide quick answers, researchers argue that it may also discourage critical thinking if used excessively for tasks that individuals can perform on their own.
To address these concerns, the study recommends designing AI systems with more awareness of long-term user outcomes. Instead of providing direct answers in every situation, such systems could guide users through problems, encouraging them to think independently while offering support when necessary.
The authors likened this approach to a mentor who helps a student understand a concept without solving the problem for them. They argue that balancing assistance with cognitive effort could help preserve users’ problem-solving abilities while still benefiting from technological support.
Although the study has not yet undergone peer review, its findings add to growing discussions about how artificial intelligence is shaping human thinking and learning habits.
