Is Your AI Use Making You Dumber or Smarter?

I use AI probably six hours a day. I have two OpenClaw agents, Codex running constantly, and Claude pumping things out. I am extremely prolific. I could not be doing half of what I’m doing without it. I’m all the way in.

I also work as an AI designer, so I see what these tools can do up close, and what they can’t. And the more I use them, the more convinced I am that the question isn’t whether you use AI. It’s how you use it. Because the same tool can make you sharper or slowly make you dumber, depending on the way you use it.

Two ways to use AI

The first way is to cognitively offload: to let AI do the work for you. If you’re in school, you say “write this paper for me,” and you turn it in without reviewing it. You never wrestle with the ideas. You never have to try.

The second way is to be empowered: to use AI as a thinking partner. You bring the question, the judgment, and the vision. You push back on what it gives you, refine it, and stay in charge of where it’s going.

When you use AI as a thinking partner, your cognitive capabilities and your creativity stay intact. When you use it to stop thinking, you start to lose them.

What offloading does to you

On my stream, a commenter joked, “I already don’t have critical thinking. Cognitive offloading is badass.” It’s funny. But there are permanent consequences to cognitive offloading. You use it or you lose it.

The research is starting to show what that looks like. In a January 2025 study in the UK, Michael Gerlich found that heavier AI use went with weaker critical thinking, and that cognitive offloading largely explained the link. The youngest participants, ages 17 to 25, relied on AI the most and scored lowest. It’s a correlation, not proof of cause, but the pattern is hard to ignore.

A 2025 study from Microsoft Research and Carnegie Mellon surveyed 319 knowledge workers who use AI every week. The more confident people were in the AI, the less critical thinking they did. The more confident they were in their own abilities, the more they questioned and improved what it gave them.

And in a small 2025 study from the MIT Media Lab, researchers led by Nataliya Kosmyna recorded the brain activity of 54 students writing essays with ChatGPT, with a search engine, or with no tools at all. The ChatGPT group showed the weakest brain connectivity, and 83% of them couldn’t quote a single sentence from the essay they’d just written.

Same tool, two outcomes

Here’s the study that makes the case for the two ways better than anything I could say. Researchers at the University of Pennsylvania, led by Hamsa Bastani, ran an experiment with nearly 1,000 high school math students in Turkey, published in PNAS in 2025. One group practiced with a standard ChatGPT-style interface that would just give them answers. Another used a version prompted to act like a tutor: it gave hints and refused to hand over solutions. A third group had no AI at all.

During practice, both AI groups crushed it. The answer-giving group did 48% better than the no-AI group, and the tutor group did 127% better. Then the researchers took the AI away and gave everyone an exam. The answer-giving group scored 17% worse than students who never had AI. The tutor group held steady, scoring about the same as the no-AI group.

Same underlying model. Completely different outcomes. The difference was whether the AI did the thinking for the students or made them do it themselves.

The MIT study found a version of this too. Students who wrote essays on their own first and then used ChatGPT to rewrite them showed more brain engagement, not less. When your brain leads and AI follows, you get the benefit without the debt.

Taste is the job

I’ve worked as a designer for a long time. I know what to look for and what to think about. When I type into Claude Design, “build me this,” it does a really good job. But it takes me about ten iterations before it produces something actually aligned with my users. Ten prompts to design an entire experience is nothing compared to how long that used to take. Could it get down to three? Probably. Can the human be eliminated? Not for a while.

I’ve watched what happens when project managers are put in charge of vibe coding solutions. The results aren’t good for users, because there are so many things a designer is trained to catch. AI also builds things in wacky ways. Just today, I asked my OpenClaw agent to do something specific, and it came back with: I couldn’t do what you asked, so I did something else. No. I asked for it a certain way because that matters.

The company I work for has fantasized about replacing employees with AI for three years, and it has only added employees. Everyone is in the building phase, and the ground keeps shifting: what took six months to build can be rebuilt in five minutes half a year later. In that world, the eye of taste, the ability to tell AI what to do and what to focus on, isn’t going away.

And here’s what worries me. The specialists who were trained before AI carry knowledge that needs to be passed on for AI output to stay good. If the next generation offloads everything, who will be left to know what good looks like? The skills that will matter most in the future workforce are critical thinking and discernment, and a lot of people won’t have them.

Decide what you won’t offload

This applies to businesses too. OpenAI and Anthropic are building models that work well for everyone, for the general public. What those models don’t have is your business knowledge: exactly what your business needs, what it values, and what makes it different.

If you type “build me a website for a doctor’s office” and hit enter, you’ll get a website for a doctor’s office. It’ll look a lot like every other doctor’s office site made with the same prompt. The value is in what only you can bring.

So start with your values. If you’re a doctor, why are you a doctor? Is it because you love helping people? Because you like connecting with them? Because you want money? Whatever the answer, it tells you what you will not offload onto AI and will keep for yourself, and what you can hand off because it doesn’t matter as much to you or your business. Then figure out what knowledge your business needs to feed the AI so it can actually do the work your way.

For the next five to ten years, I think capturing that kind of business knowledge and making it meaningful to AI is going to be one of the biggest wins out there.

Smarter or dumber is a choice

I’m not telling you to use AI less. I use it more than almost anyone I know. I’m telling you to use it on purpose.

Think first, then ask. Push back on what it gives you. Check its work, especially where you know your stuff. Keep doing the hard parts that make you who you are. Treat it like a brilliant collaborator, not a replacement for your own mind.

AI can make you dramatically more capable. It can also quietly make you less capable while you feel more productive than ever. The tool doesn’t decide which one happens. You do.

Sources

Thank you for raising the vibration of earth. If this speaks to you, join UNICULT.

What's aligned