CEOs are warning that AI without hard thinking just gives you a faster version of yesterday. The tool doesn’t build judgment. The reps do. A discipline-of-thought episode.
A couple of weeks ago, the CEO of one of the biggest retailers in the world stood in front of a room of executives and said the biggest risk with AI is ending up with “a better version of yesterday.”
Her name is Malina Ngai, she runs AS Watson Group, and she wasn’t saying AI is bad. She was saying most companies are using it to make the same old process run faster. Same decisions, same blind spots, same weak spots. Just quicker.
The number that backs her up is rough. According to the Fortune piece where she said it, only about 5% of companies that have rolled out AI agents actually restructured how they work. Everyone else bolted a jet engine onto a horse cart and called it transformation.
I run a web agency. I use these tools every day. I also run a gym. And I’ve watched the same thing happen in both places for years, long before AI showed up. Somebody grabs a tool hoping it’ll do the hard part for them. It never does.
Speed multiplies whatever you already are
That’s the whole thing in one sentence. If you’re a sharp thinker, AI makes you sharper. If you’re sloppy, it makes you sloppy faster and with a lot more confidence.
I’ve watched a client want AI to fix a marketing problem that was actually a “we don’t know who our customer is” problem. The tool wrote gorgeous copy aimed at nobody. I’ve watched a guy at the gym buy the program, the app and the tracker and still skip the hard sessions. The tool didn’t make him disciplined. It gave him a nicer-looking record of not being disciplined.
The trap was never the tool. The trap is believing the tool did the thinking.
Judgment is the thing getting scarce
Another exec at that forum, Christina Zhu from Walmart China, made the point that AI makes the human more important, not less. Data can tell you what customers picked. It can’t tell you what they needed and never found. That takes a person who’s paying attention.
Judgment isn’t knowing the answer. It’s knowing which question matters, what to ignore, when the data is lying to you, and when to move before you have all of it. None of that comes from a prompt. It comes from having been wrong enough times, in enough situations, that you’ve built a feel for it.
Martial artists already know this. Anybody can learn a technique from a video. Knowing when to throw it, against this guy, in this moment, with this much gas left in the tank? That’s reps. Only reps.
How the people who make real calls actually train
Look at how special operators, first responders and fighters build judgment. They don’t train with the answer key. They train under ambiguity. Bad information, time pressure, real consequences. Then they debrief. Then they do it again.
Those people use tools constantly. Intel, comms, maps, data. But the tool feeds the decision. It never makes it. The second the tool makes the decision, you’re not a leader anymore. You’re a relay.
The crutch test
Here’s how you know which side of the line you’re on. Ask yourself: am I using this to sharpen my thinking, or to skip it?
Sharpening looks like this. You’ve already thought about the problem. You have a position. You hand it to the tool and tell it to attack it, find what you missed, or clear the grunt work so you can spend your energy on the call.
Skipping looks like this. You don’t have a position. You ask the tool what to do. It gives you something confident and plausible and you go with it, because it sounded smart and you didn’t want to sit in the discomfort of not knowing.
From the outside, both look identical. Only you know which one you’re doing.
Where AI belongs in a decision
It belongs in the research. Let it gather and summarize. That was always grunt work.
It belongs in the stress test. After you’ve made your call, let it argue with you. Great sparring partner. Terrible coach.
It belongs in the execution. Once the decision is made, let it draft and build and ship.
It never belongs in the decision itself. The moment where you weigh it, own it and put your name on it. That’s yours. Hand that over and you’ve handed over the one thing that made you worth following.
The reps that build a decision-maker
Decide before you ask. Write down what you’d do before you open the tool, then compare. Over time you’ll learn where your judgment is strong and where it’s weak.
Debrief your calls. Once a week, look back at the decisions you made. Which ones were you actually thinking on? Which ones did you just accept?
Sit in the uncomfortable part longer. Not knowing is where judgment gets built. The tool is designed to end that feeling as fast as possible. Don’t let it.
Own the outcome out loud. If it goes bad, “the AI suggested it” is not a sentence a leader gets to say.
The warrior take
The tool doesn’t build the judgment. The reps do.
A better version of yesterday is still yesterday. If you want a different tomorrow, you’ve got to do the thinking that’s uncomfortable, and nothing can do that part for you.
This week’s challenge: for seven days, before you use AI for anything that matters, write your own answer first. One line. Then ask. Then compare. At the end of the week you’ll have a map of exactly where you’ve been leaning on a crutch.
And read the Fortune piece. It’s short and worth it:
https://fortune.com/2026/09/15/ceos-ai-efficiency-trap-fortune-leaders-forum/
