The People Using AI the Most Are the Least Afraid of It.

Last week I taught an AI First product course to a room of experienced operators. Smart people, real careers, companies they built.

Afterward, three of them came up separately and said some version of the same thing. They had not started yet. They asked their question quietly, the way you ask about something you feel you should already know.

None of them said they were afraid. All of them were.

The fear is real. It is also aimed at the wrong thing, and the way out is smaller than anyone expects.

Thirty-Seven Percent

The CNBC and SurveyMonkey workforce survey published this month puts hard numbers on it. Thirty-seven percent of US workers never use AI at work. Not rarely. Never.

When people are asked why, the reasons sound reasonable and technical. Privacy and security concerns. Doubts about accuracy. Ethical questions. Those are legitimate, and I have written about all three this month.

But notice something about those answers. They are the reasons you can say out loud in an office without cost. They are professionally respectable. Nobody gets demoted for citing security.

I am not calling anyone a liar. Those concerns are genuine. I am saying they are rarely the thing actually keeping a person from opening the tool on a Tuesday afternoon.

The Number That Should Change Your Mind

Here is the finding from that same survey that reframed this for me.

Among people who use AI daily, 27 percent say they feel more secure in their job and 22 percent feel less secure. Roughly even, tilting positive.

Among people who use it weekly, only 10 percent feel more secure while 25 percent feel less secure.

Read that twice. The people who use it the least, while still using it, are the most afraid. Confidence does not come from keeping your distance. It comes from depth.

That makes sense once you sit with it. If you have used a tool twice, all you know is that it is powerful and you are not good at it. That is a genuinely frightening combination. If you have used it two hundred times, you know exactly where it is strong, where it is confidently wrong, and what it cannot do at all. Knowing the edges of a thing is what makes it stop being scary.

The same survey found 68 percent of users say AI has saved them time, rising to 84 percent among daily users, and 56 percent say they feel more confident in their own skills. Not replaced. More confident.

The fear does not resolve by waiting. It resolves on the other side of use.

What You Are Actually Afraid Of

Let me say the quiet part, because I do not think it gets said enough.

You are not afraid of the technology. You are afraid of being bad at something in public.

You have twenty years of experience. People come to you when they are stuck. Your entire professional standing rests on being the person who knows. And this thing asks you to be a beginner again, visibly, in front of people who already look up to you.

That is not a small ask. It is the hardest ask you can make of someone competent.

It also explains the pattern I saw in that classroom, which runs exactly backward from what you would predict. The most experienced people in the room had started the latest. Not because they were slower. Because they had the most to lose by being seen fumbling.

Being new at something is uncomfortable at any age. It is worse when you have a title.

The demos make it worse. Every one you have seen was somebody's best take, with a refined prompt and a cherry-picked result. Nobody films the forty minutes where the model misreads the question and you rewrite the request four times. So the first time your own result comes back mediocre, you conclude the tool is overhyped or that you are bad at it. Usually neither. The first two weeks are supposed to be unimpressive.

Start Where You Cannot Be Wrong

Here is the specific move I give people, and it solves both problems at once.

Pick a task you already know the right answer to.

Something from your own work, last week, that you already completed. Draft an email you already sent. Summarize a document you already read. Write the first version of a plan you already made.

Two things happen. First, you can actually judge the output, because you know what good looks like for this exact task. You are not trusting it blindly, which handles the accuracy concern honestly rather than by faith. Second, there is no risk. The work is already done. Nobody is watching, nothing depends on it, and you cannot embarrass yourself.

Do that five times before you use it on anything live. You will learn more about where the tool is strong and weak in one afternoon than in six months of reading about it.

What I'd Do This Week

Pick three finished tasks and redo them. Work you already completed and already know is good. Compare. Notice specifically where the output was useful and where it was confidently wrong.

Do it privately first. You do not owe anyone a public learning curve. Get your footing, then work in the open once you have something worth showing.

Write down what it got wrong. That list is more valuable than any course. It is the beginning of your own judgment about this tool, and judgment is the part nobody can hand you.

Give it thirty days before you decide. Not thirty days of thinking about it. Thirty days of touching it most days, on real work.

Say out loud that you are learning. If you lead people, this one matters most. The moment you admit you are figuring it out, you give everyone below you permission to stop hiding.

The Part That Matters

The fear is not irrational and I am not going to tell you to just get over it.

But the data is clear about where it goes. The people using this thing every day are not the most frightened ones. They are the least. They got there by starting badly and continuing anyway.

You already know how to be a beginner. You did it once, and it worked out.

Do it again, quietly, on work that is already finished. Nobody has to watch.