AI adoption is a people problem, not a software problem.

The tool is ready. The people were never asked how they felt about it.

Here is how most companies roll out AI.

Buy licenses. Send a memo. Host a lunch-and-learn with a slide titled "The Future Is Now." Check the usage dashboard a month later. Watch it flatline like a houseplant nobody waters.

Then leadership concludes the tool is not ready. A bold diagnosis from people who have not opened it since the lunch-and-learn.

The tool is ready. The people were never asked how they felt about it.

The software is the easy part

The models are good. They write, summarize, analyze, draft, and code better every quarter. Access is not the bottleneck. Almost anyone in your company can open a chat window today. Most already have. To ask it for a sourdough recipe.

So why don't they?

Because adoption is not a technical event. It is an emotional one. And the emotions are reasonable.

Because adoption is not a technical event. It is an emotional one.

What people are actually afraid of

When I talk to people one on one about AI, I hear the same three fears. Rarely out loud. Usually between the lines.

"I will look stupid." Nobody wants to be the person who types a bad prompt in front of the team. Adults hate being beginners. We would rather quietly struggle for a decade than ask one dumb question on a group call. We spent years becoming good at our jobs, and now a chat box makes us feel like interns.
"I will be replaced." If the thing can do my job, what am I for? This fear does not go away because a VP says it is "augmentation." People can smell a talking point. Usually from two floors away.
"My expertise stops mattering." The senior copywriter, the analyst with the perfect spreadsheet, the designer with fifteen years of taste. Their status came from doing the hard part. If the hard part gets easy, their status feels up for grabs.

I once fell down a rabbit hole reading about Daniel Kahneman and Amos Tversky (I have to admit, I didn't know their names and had AI look them up), two psychologists with more brains than the average college library. My favorite nugget from them: losing something stings about twice as much as gaining the same thing feels good. Your team is not weighing "I could save three hours a week." They are weighing "I could lose the thing that makes me valuable." Of course they hesitate. I would too.

What actually works

I learned all of this by doing it wrong first. The first time I posted to the group about AI, I was thrilled. I had links. I had use cases. I had an alarming number of exclamation points. The response? Crickets in an empty room. Exactly one person replied. One. I still don't know if they were excited or just being kind to me. I had made the classic mistake. I fell in love with the tool and forgot to ask how anyone felt about it. What finally cracked it was something I had not done since second grade. Show and Tell.

I fell in love with the tool and forgot to ask how anyone felt about it.

I am the person at STM whose job includes making this stick. Here is what I have learned.

Start with one annoying task, not a strategy. Ask someone what part of their week they hate. Fix that with AI, sitting next to them. One saved Friday afternoon does more than ten keynotes.
Let people be bad in private. Give them room to experiment without an audience. Confidence comes before sharing.
Make wins public, and credit the person, not the tool. "Jess cut a two-day report to two hours" builds momentum. "AI saved two days" builds anxiety. My favorite tool for this is gloriously old school: Show and Tell. Someone gets up and shows off the nifty thing they got the robots to do for them. A prompt that writes their weekly recap. A workflow that murdered a spreadsheet. A coworker's grin is more convincing than any slide I will ever make. Turns out adults still love Show and Tell. We just stopped bringing the class hamster.
Talk about jobs honestly. Say what will change. Say what will not. Vague reassurance is worse than an uncomfortable truth.
Show experts their taste got more valuable, not less. When anyone can generate fifty options, the person who knows which one is good becomes the most important person in the room.
Measure work moved, not logins. Usage dashboards measure curiosity. So does a button labeled "Do not press." Count the tasks that now run differently.

The real job

AI adoption is change management with a shinier logo and a lot more webinars. The playbook is old. Listen first. Lower the risk of trying. Celebrate the people who go first. Be honest about what is hard.

The companies that win with AI will not be the ones with the best licenses.

They will be the ones whose people felt safe enough to use them.