On campus and online, the cursor blink is a dare. You can outline a paragraph the old way. Or you can paste a prompt and let a chatbot do the heavy lifting. A new MIT Media Lab study tried to measure what that choice does to our brains by wiring up writers with EEG caps while they composed SAT-style essays with or without AI. The headline finding is simple enough to go viral. Participants who leaned on ChatGPT showed the lowest neural engagement and produced more formulaic writing than those who worked solo or used a search engine. The full picture is more layered. It is also a snapshot, not a verdict.
The setup was straightforward. Fifty-four adults were split into three groups: one wrote with ChatGPT, one with Google, one with no tools. They completed multiple sessions across several months while researchers recorded brain activity and evaluated the essays with teachers, an AI judge, and text analysis. On average, the ChatGPT group showed weaker connectivity across key networks and scored lower across neural, linguistic, and behavioral measures. The solo writers showed the strongest connectivity and the highest sense of ownership over their work.
If you are picturing a parade of copy-paste, you are not far off. Over time, many AI-assisted writers allegedly outsourced more of the task and converged on the same phrases and structures. Two teachers reportedly called the results “soulless.” Time’s write-up adds that EEG readings pointed to lower executive control and attention among the ChatGPT cohort, and that the researchers tracked activity across 32 regions to make their case. The study is a preprint, so the word “preliminary” belongs on every conclusion, but the pattern it describes will feel familiar to anyone who has read too many bot-adjacent essays.
The contrast with search is a cultural tell. The Google group still used the internet, but they had to hunt, read, and synthesize. That friction seems to matter. In the lab, search users reported higher satisfaction than AI users and looked more engaged on EEG, landing between the solo writers and the chatbot contingent. If you have stopped going to Google for first-pass questions and started “asking inside the chat,” this is a quiet reminder that how we get information shapes what we do with it.
There is also a twist that complicates the doom. In a final session, some participants switched conditions. People who had been writing without tools then tried ChatGPT, and people who had started with ChatGPT had to write without it. The solo-first group still showed strong recall and broad activation when they later added AI. The ChatGPT-first group, writing without it, struggled to remember their own essays and showed weaker memory-related rhythms. The implication is not “AI ruins your brain.” It is that habits set the tone. Start with thinking, then layer tools, and your brain stays in the loop. Start with delegation, and your brain may check out.
If you are bristling at language like “harm,” note that the project’s own page begs reporters to chill. The authors ask people to avoid scare-terms like “brain rot” and to treat the work as early evidence rather than destiny. That caveat matters because the Media Lab is trying to map where cognition gets lighter or heavier when we add AI, not to end the conversation. The paper’s limitations are spelled out, including a small, local sample and a focus on a single kind of writing task.
The anxieties floating around this study did not appear from nowhere. Earlier this year, the Media Lab highlighted research suggesting that heavy ChatGPT users report greater loneliness and dependence and have fewer offline relationships. That does not mean chatbots cause loneliness. It does mean that people who lean hard on bots for company often look lonelier in the data. Pair that with the new EEG work and you start to see a cultural pattern. Outsourcing attention is efficient. It is also sticky.
Outside education, the motivation story is getting louder. Harvard Business Review summarized work in May that found people got more productive with gen-AI help and less motivated when they turned to tasks without it. You can call that a vibes tax on automation. The output gets a boost. The appetite to grind without the boost goes down. For managers and students, the tension is the same. If AI makes the path smoother, how do we keep wanting to walk when the ground gets rough again.
This is also why institutional caution is news. In a separate episode, MIT said in May that it no longer stands behind a widely circulated doctoral preprint about AI’s productivity lift in a science lab. The school asked for it to be pulled after concerns over data and method. That withdrawal does not touch the Media Lab preprint, but it underlines how fast AI papers can move through headlines before the vetting settles. The smart posture right now is to read with curiosity and brakes.
So what is the cultural read here. The study lands in a moment when school districts flirt with “AI across the curriculum” while creators flood feeds with AI-flavored life hacks. The lead author even told Time she worries about someone green-lighting “GPT kindergarten,” and she set little traps in the paper to catch auto-summaries. Of course people still used LLMs to summarize it. That is the point. We like shortcuts. We are building our days around them. The risk is not that a model steals our thoughts. It is that we stop building the muscle to have them.
If you are a student, a parent, or the manager of a team that writes for a living, the operational takeaway is surprisingly boring. Keep human ideation and synthesis at the front of the process. Use AI to tidy, translate, and probe ideas you already own. In the Media Lab’s data, brains look most alive when the work starts in your head and tools come second. That is not purist advice. It is a workflow. If the goal is original thought, you do not want the first draft to be the last thing you remember.
The culture will not abandon AI. We are not going back to card catalogs or strict “no tech” classrooms. What this preprint offers is a signal about attention in the age of assistance. Attention is a habit. Delegation is a habit. Whichever one you practice first will teach your brain what to expect next. That is not a moral judgment. It is a daily choice that scales over time. The cursor still blinks. The dare is still there.
WhatsApp
Instagram
Threads
