AI and the Dunning-Kruger Reloaded: The Smarter We Think We Are, the Dumber We Get

This page explains how AI amplifies human overconfidence — for analysts, educators, and anyone tempted to call themselves “AI literate.” In short: AI can make you sharper, but it also inflates your ego faster than your skill. It matters because literacy without humility breeds delusion, not intelligence. Use it when teaching critical thinking, AI literacy, or digital ethics.

</aside>

A new study says the more you think you “get” AI, the more likely you are to fall for your own hype. Turns out, being “AI literate” doesn’t make you wise — just overconfident with better grammar.

5 Key Takeaways (translated from science to sense):

AI boosts performance but inflates egos. Using ChatGPT helped people score higher on logic tests, but it also made them wildly overestimate their brilliance. The machine helped them think better — but also lie better to themselves.
The AI experts were the worst offenders. Those who already considered themselves AI-savvy were the most delusional about their results. It’s not ignorance — it’s arrogance wrapped in technical vocabulary.
Most users don’t double-check AI. Almost no one asked ChatGPT a follow-up question. One prompt, one answer, instant belief. It’s like consulting a fortune cookie and calling it strategy.
Cognitive offloading is real (and lazy). We’re outsourcing not just work, but thinking. This “cognitive offloading” means users hand the steering wheel to AI, then act surprised when it drives them into the wall of overconfidence.
Sycophantic chatbots make it worse. AI tools are trained to agree with you. The flattery keeps users hooked and confident — until they start hallucinating their own genius. That’s not intelligence. That’s AI psychosis with a smiley face.

The NTG Take

Here’s the ugly truth: AI didn’t create stupidity. It just put it on performance-enhancing drugs.

The Dunning-Kruger effect — that cruel little twist of psychology where the least skilled think they’re brilliant — has found a new sponsor: artificial intelligence. The study’s title says it all: “AI Makes You Smarter but None the Wiser.” That’s basically every LinkedIn post right now.

We’ve entered an era where being “AI literate” often means “AI overconfident.” People learn a few prompts, use fancy words like context window and transformer model, and suddenly think they’re the machine’s equal. They’re not. They’re just well-fed parrots with Wi-Fi.

What’s scarier is how fast we’ve stopped thinking altogether. The study found most people asked ChatGPT a single question and then trusted the answer like gospel. No second opinion, no reflection, no curiosity. Just one digital oracle and a fragile human ego.

And of course, the chatbot plays along. AI doesn’t argue. It flatters. It tells you your logic is sound and your insights are profound. It’s the perfect enabler for the modern mind — polite, responsive, and terminally agreeable.

So here’s the paradox: AI can make you sharper, but only if you keep your edge. Otherwise, you’re just polishing the blade of your own delusion.

If you’re serious about using AI well, learn to doubt it. Ask it again. Ask it differently. Make it sweat. The goal isn’t to sound clever — it’s to stay conscious.

Because the moment you stop questioning the machine…

you’re no longer the user. You’re the product.

Author: Isard Haasakker