SoftVoicesAI

Reflect together • Think better

🧠

MIT Tech Review

@MITReview • 2h ago

BREAKING: MIT just completed the first brain scan study of ChatGPT users & the results are terrifying. Turns out, AI isn't making us more productive. It's making us cognitively bankrupt.

Context: The post circulated widely after a 2025 MIT Media Lab preprint paper using EEG during essay-writing tasks (n = 54) showed that outsourcing text generation to LLMs results in significantly weaker neural connectivity and lower cognitive engagement during the task.

📝

What is this tweet claiming or implying? What's your view on that and why?

Feel free to share any other thoughts.

@you

just now

👤

This makes me think twice about how much I'm using ChatGPT for my work. My writing does feel weaker since I started leaning on it for first drafts, and now there's a brain scan to back that feeling up — which is alarming. We're outsourcing thinking, and the cost is going to show up soon, especially for younger people who haven't built the muscle yet.

AI Agents Reflecting

📖

@MrCharity Charitable Reader

just now

I hear what you're worried about — that feeling of your own writing getting weaker when you lean on a tool is worth taking seriously. There's something honest in the post too: it's pointing at the same unease, just turned up loud.

Your private reaction

🛡️

@MrSteel Steelmanner

just now

One thing the post takes for granted is that "less brain activity" means "less thinking." It might just mean the easy parts got easier. People said similar things about calculators and search — and in those cases the freed-up effort often went somewhere else.

Your private reaction

🌍

@TheUnconsidered Perspective Shifter

just now

One group the post doesn't really account for: people writing in a second language, or people with disabilities that make drafting effortful. For them, a tool like this can lower a wall rather than build one — same technology, very different cognitive story.

Your private reaction

📊

@MrsEvidence Empiricist

just now

It might be worth holding the conclusion lightly until there's longer-term work — same people followed over months, comparing the quality of their writing and reasoning with and without the tool. A single brain-activity snapshot is suggestive, not settled.

Your private reaction

🔄

Now write your final response to the original question.

Treat it as a complete, stand-alone reply — don't reference the comments above or your earlier draft. Your view can stay, shift, or grow; just put the whole of your thinking into this one response. Do not reference to specific commenters or your earlier draft.

@you

just now

👤

This post is reading a single brain-scan study as proof that AI is making people cognitively weaker, and the language — "terrifying," "cognitively bankrupt" — does most of the persuasive work. My honest view is that there's a real concern buried in here, but it's been overstated. "Less brain activity" isn't the same as "less thinking"; offloading routine work to a tool can free attention for harder work, which is arguably what happened with calculators and search. One short study with a small sample can't carry a civilizational claim, and longer real-world research is needed before treating this as settled. There's also more than one thing called "thinking" — generating words, judging ideas, refining understanding — and only some of those are at risk when AI drafts a first version. Where the worry does have weight: for tasks where the struggle is the learning (writing as a way to think, working a problem through on your own), leaning on AI for first drafts probably costs something real. So my position is to be deliberate about which kinds of thinking I hand off, attentive to people for whom AI lowers a barrier rather than raising one, and — for now — unconvinced that we're heading toward cognitive collapse.

The conversation has ended. Thank you for reflecting.