The Oxford word of the year in 2024 was “brain rot,” due to a 230 percent increase in usage of the term between 2023 and 2024. The moniker reflected a collective sense – already on the rise in 2023 – that our heavy use of digital technology has corroded our ability to think.
Now, in the futuristic era of artificial intelligence known as 2026, brain rot has acquired a whole new level of relevance. Now that it’s possible to outsource complex questions or research problems to a tool that summons answers in seconds, seemingly siphoning them out of mid-air, we need to take stock of how reliance on these digital genies blunts our own mental sharpness.
Well, a recent study presented at the Conference on Language Modeling confirms our worst fears about how using AI may be diminishing our own brainpower. The paper – titled “AI Assistance Reduces Persistence and Hurts Independent Performance” and authored by researchers from MIT, Oxford University, Carnegie Mellon, UC Berkeley, and the University of California, Los Angeles – found that spending a mere 10 minutes relying on an AI tool noticeably impairs one’s attention span and persistence on challenging tasks.
“The best collaborators maintain a balance between helping and fostering autonomy; they know when not to help,” the study’s authors wrote.
“Current AI assistants are a stark contrast to this dynamic. They rarely refuse to help, and provide instant answers to almost any query. … In this sense, AI systems are … extraordinarily helpful in the moment, but indifferent to what that help does to the person receiving it over time. In this paper, we investigate the consequences of this misalignment.”
To examine AI’s impact on human cognitive abilities and habits, the researchers conducted a series of experiments. The first one involved dividing 354 participants into two groups tasked with solving 15 fraction problems. The first group had to solve those problems on their own, while the second group could use AI for help. After 12 problems, researchers removed the AI assistance from the second group.
As expected, the AI-boosted group had more accurate answers at first, compared to the group relying on only their own gray matter. But once the AI tool was removed, the group struggled to answer questions accurately.
Similar results arose from an experience with 667 participants. The group that started out with AI assistance would simply give up or provide incorrect answers once that crutch was removed, while the group that hadn’t had AI assistance at any point rode the problems out to the end and ultimately achieved greater success.
Switching to a different kind of problem-solving didn’t make a difference; when researchers ran a third experiment with 201 participants involving SAT reading comprehension questions, the same thing happened. Participants who had come to rely on AI assistance buckled once that support was removed – both their persistence and their accuracy fell.
As the UC Berkeley press release on the study noted: “The findings backed up what anecdotes and surveys have increasingly found in recent years: An overreliance on AI for answers erodes our cognitive fitness. When AI produces answers in seconds, we get used to that speed; anything that takes longer feels inefficient.”
