Bionic brain rot?
New research suggests that AI can compromise cognitive function and problem-solving abilities in as little as ten minutes.
While the long-term effects of AI have yet to be established, a new study reveals that the technology can seriously impair brain performance in a relatively short period.
Conducted by a team of researchers from powerhouse universities — Carnegie Mellon, Oxford, MIT and UCLA — the study evaluated the effects of AI on participants using a fraction-based math test.
Half of the participants were charged with solving the problems without AI, while the other half were given access to an AI assistant for approximately 10 minutes that was then removed for the test’s final three problems.
Predictably, the AI-assisted group performed better for the first portion of the experiment. But when AI was taken out of the equation —pun intended — performance dropped dramatically.
“We find that AI assistance improves immediate performance, but it comes at a heavy cognitive cost,” said the study authors. “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.”
Indeed, both groups — the one that previously had AI and the one that never did — were both operating without the technology, the solution rate in group that had previously used AI was 20% lower.
When AI was removed from the assisted group, participants were twice as likely to skip questions as those without access to AI.
Researchers used a similar format to test reading comprehension and found parallel results, although AI assistance did not translate into stronger performance in the initial portion of the test.
Even though participants had access to AI support for a scant 10 minutes, it clearly compromised their ability to rely on their own comprehension and problem-solving skills,
However, the study authors noted that how AI was utilized had a significant bearing on the drop in solve rate and the increase in skip rate.
Sixty-one percent of AI-assisted participants reported asking AI directly for solutions to the problems. Yet the remaining members of that group who used AI for less direct support, such as hints or clarification, did not experience the same decline in solution rate.
The study authors maintain that this is evidence that AI is not harmful to cognition in all capacities, but that absolute reliance on it hinders problem-solving capabilities.
“Just 10–15 minutes of AI interaction can result in significant impairments in independent performance and persistence – capacities that are foundational to life-long learning,” they said.
“If brief exposure produces measurable erosion, the cumulative effects of daily AI use over months or years may be profound and difficult to reverse.”
A survey last year revealed that 56% of US adults use AI tools, with 28% using them at least once a week.
Previous research has suggested that AI could contribute to a “dementia crisis” because it weakens the brain systems responsible for curiosity, attention, high-order reasoning, and executive function, among other duties.
Critics say many innovations over the years have changed the way we think, including calculators, GPS, and smartphones, but nobody blamed calculators for a dementia crisis.
However, according to study authors, “current AI systems represent a new kind of cognitive scaffold: one that solves anything, rarely refuses to help, and delivers answers instantly.”
