How to Use AI Without Losing Your Ability to Think

Used as a shortcut, AI erodes your thinking. Used as a tutor, it sharpens it. Most leaders only practice the first mode.

I was scrolling headlines on my phone. Three stopped me cold. The New York Times: “Will A.I. Make Your Brain Lazy? Here’s What the New Research Actually Shows.”¹ The BBC: “AI chatbots could be making you stupider.”² India Today: “AI is slowly making your brain lazier and more dependent on it, study finds.”³

I wondered if they were right. I use AI every day. So does nearly every leader I know. If these tools were quietly dulling the judgment that built our businesses, no productivity gain would be worth it.

So instead of forwarding the scariest headline, I read the research behind it. The Times pointed to math tutoring sessions studied by UC Irvine and McGraw Hill, which found students leaning on AI to do the work for them more often after ChatGPT arrived.⁴

But a Middlebury study told a more interesting story. Students who used AI as a shortcut—handing it the whole assignment—scored worse on later tests. Students who used it like a tutor, asking it to explain and clarify, kept their strong scores and wrote better essays.⁵

The headlines flattened all of that into one scary claim. The research draws a finer line: used as a shortcut, AI erodes your thinking; used as a tutor, it sharpens it. That line is what I want to explore with you this week.

If you’re going to use AI without losing your ability to think, you’ll need to learn how to make it work like a tutor instead of a shortcut. Here are three ways to use AI that leverage your thinking instead of replacing it.

Agents take your thinking at setup and then run alone. Everything below is for generative sessions, where you stay in the conversation.

1. Ask AI to Interrogate Your Thinking, Not Finish It

Say you run a forty-person services company and you’re deciding whether to raise prices. The shortcut is typing, “Write me a pricing memo.” You get a polished page in seconds, and you’ve thought about almost nothing.

The tutor move flips the assignment. Paste in your rough notes and say, “Ask me one question at a time about this decision. Don’t write anything. When my reasoning is complete, tell me what still looks thin.” Then answer in your own words. By the end of the session, you know what you actually believe.

Close the session with one more request: “Summarize my position in three sentences.” If you don’t recognize your own thinking in those sentences, you aren’t finished. Use this before any memo, hire, or strategy decision that matters. The tool will happily draft the document afterward, and now it has something real to work from.

The UC Irvine and McGraw Hill math research puts the cost of skipping this step right in its title: “Faster Completion, Less Learning.”⁶ Researchers call it cognitive offloading, handing the mental work to a tool. Offloading a calendar invite is smart. Offloading the reasoning behind a pricing decision is how you lose the ability to make the next one.

2. Ask AI to Argue Against Your Thinking, Not Confirm It

In my experience, chatbots lean toward agreement. Ask, “Is this a good plan?” and you’ll hear some version of yes. That feels great and teaches you nothing.

Give it a decision and a role instead: “You’re a skeptical board member who has watched plans like this fail. Make the strongest case against mine. Name the three assumptions I’m most likely wrong about. Don’t offer alternatives until I’ve answered each objection.” Then write your answers. Some objections will be wrong, and rejecting a bad objection for a good reason sharpens your judgment as much as accepting a good one.

Run it this week on something real: a price change, a key hire, an acquisition offer. Then ask for the objection you’d least like to hear. That one usually points at the assumption you’ve been avoiding, and a skeptical board member who never gets tired is cheap to hire.

Zara Contractor and Germán Reyes ran an experiment on the learning impact of generative AI and separated two kinds of use. Augmentation means asking the tool to explain and clarify. Automation means asking it to generate the text.⁷ Same tool, different habit, different result.

That is the same tutor-versus-shortcut line the Times reported: you defend your position or you revise it.

3. Ask AI to Stress-Test Your Thinking, Not Replace It

Write the plan first. Every word of it. Then hand it over for a pre-mortem: “It’s twelve months from now and this plan failed. Give me the five most likely reasons, ranked. For each one, tell me the earliest warning sign I’d see.” You decide which risks are real. You rewrite the plan yourself.

The order matters. AI critiques a finished draft far better than it improves a blank page you never filled, and a draft you wrote gives you something to defend.

When the stress test comes back, sort the risks into three piles: real and urgent, real and later, and noise. Then pick one action for the first pile and put it on your calendar.

Grace Liu and her coauthors found that AI assistance reduced persistence and hurt performance when people later worked on their own.⁸ Hand a hard problem to AI often enough and you quit wrestling with it sooner. Stress-testing runs the other direction. You built the plan, so you absorb the hits, and you keep going. Persistence stays with you.

If you want another example of reading past the scary headline, see why the AI doomsday story isn’t what it seems.

Three moves. Let AI question your reasoning before it writes a word. Let it argue the other side before you commit. Let it attack your finished plan before the market does. In each one, you do the thinking and the tool pushes back.

Imagine using AI every day for the next ten years and finishing with a sharper mind than you started with. The research says that outcome is on the table.

Which of these three moves will you try in your next AI session?

Comments

If you have a question about using AI as a tutor instead of a shortcut, click here to send me an email. I read every one. Seriously. Your experiences help me write better content, and sometimes the best insights come from readers like you. 

Transforming AI from noise to know-how,

P.S. Consider the AI Business Lab Mastermind: Running a $1M+ business? You’re past the startup chaos but not quite at autopilot. That’s exactly where AI changes everything. The AI Business Lab Mastermind isn’t another networking group—it’s a brain trust of leaders who are already implementing, not just ideating. We’re talking real numbers, real strategies, real results. If you’re tired of being the smartest person in the room, this is your new room. 👉🏼Learn more and apply here.

---

REFERENCE

1. Juliana Castro Varón and Dylan Freedman, “Will A.I. Make Your Brain Lazy? Here’s What the New Research Actually Shows,” New York Times Magazine, September 29, 2026. ↩︎
2. “AI Chatbots Could Be Making You Stupider,” BBC Future, April 17, 2026. ↩︎
3. “AI Is Slowly Making Your Brain Lazier and More Dependent on It, Study Finds,” India Today, April 17, 2026. ↩︎
4. Sina Rismanchian et al., “Faster Completion, Less Learning: Generative A.I. Reduced Study Time on Math Problems and the Knowledge They Build,” arXiv:2605.21629. ↩︎
5. Castro Varón and Freedman, “Will A.I. Make Your Brain Lazy?” ↩︎
6. Rismanchian et al., “Faster Completion, Less Learning.” ↩︎
7. Zara Contractor and Germán Reyes, “Experimental Evidence on the Learning Impact of Generative AI,” arXiv:2607.08849. ↩︎
8. Grace Liu et al., “AI Assistance Reduces Persistence and Hurts Independent Performance,” arXiv:2604.04721. ↩︎

How I use AI in my writing process →