You can keep using ChatGPT and still keep your mind sharp. The trick, according to an MIT professor, is deciding whether the chatbot does your thinking or helps you do it.
That's the core of an argument from MIT professor Eric So, reported by Fox News. So argues that overreliance on AI can weaken memory, problem-solving, and foundational mastery. His reasoning is that AI removes the mental "friction" the brain needs in order to learn.
Here's what that means, why it happens, and what to do about it.
Why your brain is easy to hook
So's argument, per Fox News, is that AI takes advantage of two natural tendencies: the brain's urge to conserve mental effort, and its pull toward instant gratification.
That combination can create a feedback loop. A chatbot gives you an answer immediately, which feels good. Next time, you reach for it again, even for something simple. Over time, So argues, people turn to chatbots for the simplest of tasks.
Nothing about this requires laziness as a character flaw. Saving effort is what brains do. A tool that makes saving effort effortless is simply very easy to lean on.
The hidden cost of instant answers
Think of learning like lifting weights. The resistance is the point. When you wrestle with a math problem, try to recall a fact, or stare at a blank page, the struggle is doing the work.
So's concern, as Fox News reports it, is that an instant answer skips that struggle. You get the result without building the skill underneath it. The areas he says can weaken are memory, problem-solving, and foundational mastery, the basics that everything more advanced rests on.
The tricky part is that the output can look fine. The essay is written, the code runs, the summary reads well. The cost shows up later, when you have to do something without the tool and find the skill isn't there.
Answer machine vs. tutor
So doesn't argue that AI is simply bad. Per the report, he says it's harmful when used as an answer machine but valuable when used as a tutor or cognitive trainer.
The difference is who does the effortful part.
- Answer machine: You ask, it produces, you copy. Your brain stays idle.
- Tutor or trainer: You try first, and the AI pushes, checks, and explains. Your brain stays engaged.
Same tool, opposite effects. What changes is the order of operations and who is working hardest.
Practical ways to keep the friction
So's framing points toward habits. These are my own suggestions for applying it, not his specific prescriptions.
Try first, then ask
Before you open a chatbot, spend a few minutes attempting the problem, outlining the essay, or recalling what you know. Even a rough attempt means the AI is responding to your thinking instead of replacing it.
Ask it to quiz you
Instead of "explain this chapter," try "ask me five questions about this chapter and tell me what I got wrong." Retrieving information from memory is effortful, and that's exactly the friction that helps learning.
Request hints, not solutions
When you're stuck on a problem, ask for the next step or a nudge in the right direction. Then try again yourself. You can always ask for more help if the hint isn't enough.
Get feedback on your draft
Write your own version first, then ask the AI where it's unclear or weak. You do the creating, and it plays editor.
Save the autopilot for tasks you don't need to master
Not everything is worth learning deeply. Formatting a spreadsheet or rewording a routine email is different from a skill you need to own. A useful test: will I need to do this without help someday? If yes, keep the friction in.
Why this matters now
AI tools are becoming default helpers for writing, schoolwork, research, and decision-making. That shifts the question from "Can AI do it?" to "What does AI do to us?" It's an especially live issue for students, workers, and creators who feel pressure to use AI constantly.
So's argument is one professor's view, relayed by Fox News, and it doesn't put a number on how much any individual's skills might slip. But the underlying logic is easy to test in your own life. Notice the last time you reached for a chatbot before you'd tried anything yourself. Did you need it, or was it just easier?
The takeaway
Next time you open a chatbot, make your first move before it makes its first move. Try the problem, draft the outline, or guess the answer, then let the AI challenge you instead of replace you.
