An international team of researchers found that even brief reliance on artificial intelligence can undermine persistence and reduce people’s ability to solve problems on their own once the assistance is removed, according to The Independent.
“Boiling frog”
In experiments involving mathematical reasoning and reading comprehension, participants who used AI for 10 minutes performed worse and gave up more often when the tool was taken away compared with peers who received no help, the team from the University of Oxford, MIT, UCLA, and Carnegie Mellon reported. The researchers described these outcomes as reduced persistence and impaired unassisted performance, warning that short-term gains from AI can come at a heavy cognitive cost if they accumulate over time.
They raised alarms about a gradual “boiling frog” erosion of human cognition as everyday use of AI accelerates.
The study’s authors cautioned that skills such as fraction arithmetic and reading comprehension may appear delegable to tools, but conceptual mastery of them underpins higher-order capabilities including algebra and critical reasoning.
Their concern centers on how regular outsourcing of effort could dull the motivation and stamina required for long-term learning, with small degradations compounding until they become difficult to reverse.
“Desirable difficulties”
Grace Liu, a co-author from Carnegie Mellon University’s Machine Learning Department, said the issue is not that AI makes people less intelligent in a blunt sense, but that it can quietly strip away the difficulties that help build durable competence.
“The concern is about what cognitive scientists call ‘desirable difficulties’ - the productive struggle that builds skill over time. If AI routinely removes that struggle, people may get the right answer in the moment, but develop less robust independent capability,” she said.
Liu added that the scale and scope of the effect still require investigation. “It’s not about AI making us ‘dumber’ - it’s more subtle than that. But how significant this effect is at scale, and across different contexts, needs more research,” she added. “It’s not a reason to avoid AI, but it is a reason to design and use these tools carefully,” Liu concluded.