AI 'Writing' Is Mind-Numbing Digital Dookey

I came across a study last week that made me put my phone down and think.

Before I get into it, let me be honest: I use AI tools every day. ChatGPT for random 2am questions about science, photography, history, theology. Grammarly for editing my notorious addiction to typos. Claude for managing large writing projects. Perplexity for deep research that requires long study sessions and credible citations. Leonardo.ai for images. Google Gemini for tasks across all my Google suite.

These tools are powerful. They’re also powerfully insufficient for anything requiring genuine human creativity, thinking, or originality. They excel at fast recall across vast information streams. They’re far less useful for producing things that haven’t been produced yet. In my experience, they generate clichés more than innovations.

I view them as tools that should enhance human talent, not replace it. Maybe that’s me being overly precious about old-school work discipline versus forward-thinking tech optimism. But I can’t shake a recent conversation with a friend who told me she was letting go of her communications contractors because she could “just drop stuff in ChatGPT for everything.” No shade, but that’s gross and alarmingly amateurish.

Why most AI writing is terrible (and disrespectful)

If AI makes people think they can replace writers, they’re underestimating how dreadful their AI writing is. Much of what I see from non-writers who are suddenly “more productive” is dreadful crap, basically AI vomit wasting my time. It’s disrespectful to publish drivel without editing beyond ChatGPT’s first output. When I read their work, I’m offended. Seriously. I know quickly that I’m not reading their actual reasoning and thinking. I’m reading their overly clever Google searches written by third-grade AI tools.

We all deserve better than lazy digital dookey wasting our precious reading time.

The tell-tale signs of AI writing are flooding my inbox. I see how quickly it’s reducing the cognition of people who rely on AI to be their brain, voice, and thinking. It’s scary that this is being normalized in professional places, devaluing professionalism and putting lazy minds on par with thinking writers.

All that said, I still use these tools with a bit of wonder. I’m trying to learn as many as possible. I see it as an emerging demand that will leave me behind if I don’t recognize it. I’ve been disappointed by their many shortcomings even as I’ve been amazed by their power to save time on basic tasks. Sometimes I finish work much faster because of them. Other times I waste hours trying to get AI to produce something original.

The MIT study: How ChatGPT changes your brain

Which brings me to this MIT study. Researchers wanted to know what happens in our brains when we use AI to help write essays. What they found is both fascinating and unsettling.

They recruited 54 people and split them into three groups. One group wrote essays using only their brains, no tools at all. Another group could use Google to search for information. The third group could use ChatGPT (technically GPT-4o, the AI writing assistant). Each person wrote three essays over four months, always using their assigned method.

In a fourth session, they switched things up. People who’d been using ChatGPT had to write without it. People who’d been writing on their own got to try ChatGPT for the first time.

The researchers weren’t just reading the essays. They were measuring brain activity using EEG (electroencephalography). They put caps with sensors on heads to detect electrical signals showing which parts of the brain were working and how those parts were communicating with each other.

What they found: brains of people writing without any tools showed the strongest, most widespread activity. The search engine group showed moderate brain activity, somewhere in the middle. But the ChatGPT group? Their brains showed the weakest connectivity, especially in parts responsible for deep thinking, working memory, and creative problem-solving.

Brain connectivity is basically how different parts of your brain talk to each other. When you’re really thinking hard, wrestling with ideas, trying to figure out how to say something, your brain lights up with activity across multiple regions. That’s your brain doing the work of learning and creating.

When people used ChatGPT from the start, that connectivity dropped by up to 55% compared to people writing on their own.

Even more interesting: the researchers introduced the phrase “cognitive debt.” It’s like financial debt, but for your brain. You’re borrowing thinking power from the AI instead of building your own. In this way, AI is a payday loan vendor for cognition.

And just like financial debt, there are consequences. When the ChatGPT users were asked to quote something from the essay they’d just written, 83% of them couldn’t do it. Not one accurate quote from their own work. Meanwhile, almost everyone in the other two groups could easily quote what they’d written.

That’s not just forgetting a detail. That’s a sign the ideas never really became theirs. Their brains never took ownership of the thinking.

The fourth session revealed something even more concerning. Remember how they switched the groups? People who’d been using ChatGPT for three sessions and then had to write without it showed weaker brain connectivity than people who’d been writing on their own all along. Even without the AI, their brains seemed different. Less engaged. The researchers described it as “under-engagement of alpha and beta networks,” which are brain wave patterns associated with focused attention and active thinking.

It’s like their brains had gotten used to not working as hard.

The right way to use AI for writing

This all sounds bleak, I know. But there’s a hopeful flip side. People who wrote on their own first and then tried ChatGPT in the fourth session showed spikes in brain activity across all frequencies. The researchers think this happened because they were doing extra mental work, comparing what the AI suggested to what they’d already been thinking about. They were integrating the AI’s ideas with their own, not just accepting them wholesale.

The study also looked at the essays themselves using natural language processing, which analyzes patterns in writing. The ChatGPT group’s essays were more homogeneous (they all sounded similar to each other), used more named entities (specific people, places, things), and often had weird third-person phrasing that AI tends to favor. Human teachers described many of these essays as “generic” and “soulless.”

Here’s what I’m taking from this: the tool matters less than the sequence. If you start with your own thinking, struggle with ideas, make the mental effort to figure out what you want to say, and then use AI to help refine or expand, your brain stays engaged. The neural pathways keep firing. You’re building cognitive muscle.

But if you start with AI, if you outsource the hard thinking from the beginning, something changes. Your brain doesn’t build those connections. The ideas don’t stick. You end up with work that might score well but doesn’t feel like yours because, neurologically speaking, it isn’t.

What cognitive debt means for education

The researchers are careful to note their findings are wholly preliminary. The study hasn’t been peer-reviewed yet, so other scientists haven’t tested their work to see if it’s solid. They had a relatively small number of participants. There are questions about whether this applies to other kinds of writing or other uses of AI. But the pattern they found is consistent and measurable.

It also aligns with my vibes on the matter. Unscientific, perhaps a bit grumpy, but I trust my intuition on this.

I’m obsessing about what this means for learning. I’ve watched educationists chase shortcuts, trying to make things easier and faster, with or without success. I’ve been guilty of it myself, looking for ways to streamline work that maybe shouldn’t be streamlined. But your brain doesn’t learn by watching someone else think. It learns by doing the thinking itself. That struggle, that effort, that frustration when you can’t quite find the right words? That’s not a bug. That’s the way. That’s how neural pathways get built.

The “just drop it in ChatGPT” approach’s problem isn’t that the output looks good on the surface. It’s that nothing is happening in your brain when you produce it, making me wonder which of you—the “writer” or the AI—is the robot. You’re not wrestling with the ideas you’re publishing. You’re not making connections. You’re not building the cognitive muscle that makes you a better thinker over time. You’re outsourcing the exact activity that would make you smarter.

How to use AI tools without losing your mind

AI tools like ChatGPT aren’t going away. And they shouldn’t. They’re powerful, useful, sometimes revolutionary. But this study suggests we need to be thoughtful about how and when we use them. Starting with our own thinking matters.

The cognitive effort matters. The struggle matters. We can’t outsource learning.

We can only support it or undermine it. Right now, the evidence suggests that starting with AI might be undermining something essential about how we build knowledge, form memories, and develop ideas we actually own.

That phrase, “cognitive debt,” is sticking with me. We’re borrowing against our own minds. And unlike money, you can’t just pay this debt back later. Those neural connections either form or they don’t. That thinking either happens or it doesn’t.

The good news is we get to choose.

Every time we sit down to write, to learn, to think through something difficult, we decide whether we’re going to do the work or outsource it. This study is a reminder that the choice has consequences we can actually measure in our brains.

I’m still figuring out what this means for how I use AI. I’m not abandoning these tools. They’re too useful for certain tasks, and the demand to understand them isn’t going away. But I’m more aware now of when I’m using them to enhance my thinking versus when I’m using them instead of thinking.

In the end, it’s not just about my output. It’s about what happens inside my head while I’m getting there.