
Experts With Power Tools, Not Monkeys With Machine Guns
Supervisors in a Leadership Quarterly study trusted the same work 68 percent of the time when it was done alone and 24 percent when it was done with AI. AI can be a power tool or a machine gun, and which one you get depends on who is holding it.
Jarrett Stanley·Chief Marketing Officer, Nationwide Mortgage Bankers
5 min readNovember 18, 2026
On the last episode of The TrueTone Show, Corey Shelton asked me to say something twice.
He had just said he wants his team using AI for everything it helps with, but he does not want people who let AI tell them what to do. I said you want experts with power tools. What you do not want is a monkey with a machine gun.
(For the record, we do not advocate handing a firearm to any primate.)
The line landed, I think, because it describes one tool in two very different pairs of hands. A carpenter with a power tool builds the house faster. A monkey with a machine gun fires a lot of rounds, nobody can tell you where they went, and each one causes more damage.
AI can be either one. Which one you get depends on who is holding it.
Your boss is already reacting to this
Researchers at the Hebrew University of Jerusalem recently published a study in The Leadership Quarterly that put people in the role of supervisor. They trusted an employee who did the work alone 68 percent of the time. They trusted an employee who did the same work with AI 24 percent of the time.
Same work, about a third of the trust.
I understand the instinct, even though I think it is wrong. On the surface, if you do not know what you are looking at, the expert and the monkey turn in work that looks about the same. The supervisor cannot see the difference, so they discount everybody.
The same study points at a fix. When the company openly encouraged AI use, the penalty shrank until it was no longer statistically significant. That makes a lot of the trust gap a policy problem, and leaders can close much of it by saying out loud that they want their people using these tools.
I want my team using AI anywhere it makes sense for the task. I am not in the business of paying more for slower work.
What the expert brings
I can tell the difference between a $10 bottle of wine and a $20 one. I cannot tell you the difference between a $20 bottle and a $200 bottle. I do not have the palate. A sommelier who has tasted thousands of wines could tell you in seconds, and tell you why.
Websites are my $200 bottle. I have built hundreds of them, and I have opened Photoshop more than any other program on my computer. I know why margin and padding work. I know why some animation timings feel right, and others feel a few milliseconds off, and why one drop shadow looks clean while another looks cheesy. When a client tells me a page "doesn't pop," I know they usually mean the contrast is off relative to the hierarchy of the page.
I know those things because I have done the work, a lot.
That is what the expert brings to the power tool. When they ask AI for something, they are not dropping a prompt into a black box and taking whatever comes out. They can scrutinize it, validate it, and tell you how the sausage got made. The tool makes them faster at work they already know how to judge.
The monkey gets volume. Volume without judgment produces fool's gold: products that look great on the surface and are busted underneath. The uncomfortable part is that the people approving the work often cannot tell the difference either. Everybody celebrates the launch, and the cracks show up when real people start using it.
The valuation issue
Before AI, if you wanted a raise, you leveled up. You earned a certification, took on more complex projects, and did more. Your boss could see the growth.
Now an expert shows up with a bigger, better body of work, and there is a real risk the response is: I don't need to give you that raise. You just asked AI. I can get anybody to do that.
Meanwhile, that same boss is comparing the team against shops that use AI and charge less. You cannot have it both ways. You cannot discount someone for using AI and then measure them against people who use it.
I think this is going to be a hard problem for a few years. A lot of companies will struggle to tell an expert who uses AI from someone passing off AI output as expertise. The ones that sort it out first will keep their best people, because experts who feel undervalued go somewhere that can tell the difference.
The quiz is still in person
When I was in school, we were told to learn our equations because we would not always have a calculator with us. That turned out to be wrong. We all carry one now.
AI is with us all the time too. That does not mean it belongs in every room.
If AI did your homework, you still get a 100 on the homework. Then you fail the quiz, and the quiz is in person. It is the meeting where someone asks you to defend the recommendation, explain the number, or say why this campaign and not the other one.
Corey had the best version of this on the show. Your CEO walks up and asks you a brand marketing question. Do you pull out your phone and ask ChatGPT while they stand there? Of course not. They expect you to know, because that is why they hired you.
If you do pull out the phone, the CEO's next call is to the CFO, to find out how much you make.
In a world where you can get an answer to anything, you are still the moat.
What leaders can do about it
For leaders, this comes down to four things.
Say it out loud. If you want your people using AI, put it in writing. In that study, open encouragement is what made the trust penalty stop being significant.
Grade the quiz, not the homework. When someone brings you AI-assisted work, ask them to walk you through why it works. An expert can. That conversation tells you more than the deliverable does.
Pay for judgment. Output gets cheaper every month. The ability to look at output and know what is wrong with it does not. If your raise conversations only count deliverables, you are about to underpay your best people.
Make the newer people go one layer deeper. Corey's advice for people early in their careers is to use AI for everything, then ask it why the output works. That is how you start building a palate before you have years of reps.
The episode
We got into all of this on Episode 2 of The TrueTone Show, along with whether it matters that AI made something and why taste might be the job now.
After you listen, try one thing afterward. Pick one piece of AI-assisted work on your desk and explain to someone, with the tool closed, why it is right. If you can, you are holding a power tool. If you cannot, you found out before your CEO did.
And if you have figured out how to tell the difference on your team, or how to get paid for being the expert, tell me in the comments on the LinkedIn edition of this issue.
More from The Signal

The Future Belongs to Translators
Modern work runs on three languages -- creative, finance, and code -- and the companies that win aren't the ones with the loudest specialists. They're the ones with people who can move meaning across those borders without it getting mangled.

The Cost of Noise
Noise rarely shows up as a line item. It shows up as lost judgment, rework, slow decisions, and half-built initiatives. Here is where the cost actually hits and the system-level filters that reduce it without slowing growth.

Busy Is Not the Same as Effective
A full calendar and a steady stream of deliverables can still leave you with the same outcomes at month's end. Here is how to shift from output to influence with a practical framework that builds leverage instead of burnout.
Frequently Asked Questions
Want to Build Systems Like These?
Book a strategy session to discuss how operational clarity and AI-driven marketing can transform your results.
Book a Strategy SessionCut through the noise. Subscribe to The Signal.
A weekly newsletter on AI and mortgage marketing, written by a CMO who builds with it every day.

