The Thing I Don't
Give Over to AI

The one thing I don't hand over to AI is my thinking.

I use AI constantly, from running Claude Code in my terminal to build custom solutions for clients, to drafting work, to thinking out loud about a hard problem. Claude Code has become a real part of my practice, and a lot of what sits on my laptop right now exists because AI helped me build it. That side of AI is real and it matters, and I'll write about it another day. This post is about a different use, one that sits alongside the building work and has changed my thinking as much as anything else I've learned to do. I use AI as a thinking partner, the conversational corner of AI where I bring a half-formed idea and leave with a sharper one. I talk, it talks back. I push, it pushes back. Somewhere inside that loop, my own thinking gets sharper and more honest, and that is the use I can't imagine working without.

Here is what I ask it to do. I ask it to point out my blind spots. I ask it to red team my strategy. I ask it to stress-test an idea before I bring it to a client. I ask it to poke holes in a pitch I'm too close to. I ask it to push back when I'm getting confident in a take that might not deserve that much confidence. I ask it to pull references I would not have surfaced on my own, to walk me through an unfamiliar topic, to show me the second-order effects I missed on the first pass. It is a genuinely good partner for that kind of work, and I would feel slower and less sharp without it.

But there is a real difference between using AI to sharpen your thinking and using it to skip the thinking part. The first one leaves you stronger. The second one leaves you with output you cannot defend, explain, or improve on when reality starts pushing back.

Researchers are starting to put a name on the second one. Some call it cognitive debt. Others call it skill atrophy. The World Economic Forum ran a whole session on it this year. The core finding, across a handful of studies, is surprisingly consistent. When people fully hand a task to AI, the output often looks fine in the moment. What's missing is the underlying muscle. The person who would have wrestled with the problem, and come out the other side with real understanding, did not wrestle with it. They chose from a menu. And choosing from a menu builds a very different kind of capability than making the thing yourself.

For a business owner, that bill comes due in a specific way. The work that AI most wants to do for you is often the work that builds your judgment. Pricing decisions. Positioning decisions. Hiring decisions. The hard customer conversation you've been avoiding. Those are exactly the decisions where you grow as a leader, and they are also exactly the decisions AI will happily generate a clean, confident, plausible answer for, inside of thirty seconds.

A plausible answer is not the same as a right one. And judgment, in the end, is basically the whole job.

Here is where I've landed on my own practice. You now have access to intellectual firepower that was unimaginable five years ago, available on any topic, at any hour. That is an extraordinary resource, and using it to grow your thinking is one of the best things you can do with it. What I try not to do is use it in place of my own thinking, which is a different thing entirely.

So I let it help me get smarter about a problem. I don't let it decide what the problem actually is. I let it pressure-test a decision after I've done the hard thinking on it. I don't let it make the decision in the first place. I treat it like a sharp colleague who is endlessly available, occasionally brilliant, and sometimes confidently wrong. That's exactly the kind of colleague who makes you better, as long as you stay awake in the conversation.

The bigger point, if you're running a small business and trying to figure out how much of this to invite into your week, is that AI is most valuable when it is pushing your thinking, and least valuable when it is replacing it. The first one compounds. The second one hollows you out, slowly, in a way you probably won't notice until the moment you need the muscle and it isn't there.

Use it to get sharper. Don't use it to skip the work that makes you sharper in the first place.