Corporate training existed to fix information asymmetry. AI is the first technology you can ask how to use itself. What’s scarce now is not information. It’s agency.
Every enterprise right now is building an AI adoption plan, and at the center of every plan is a training program. Curricula, workshops, certification tracks, prompt engineering courses. An entire consulting industry has formed around teaching companies to teach their people AI.
I never ran one. The best AI usage I have seen came from people who were never trained, and I have come to believe the training instinct is not just unnecessary. It is solving the previous era’s problem.
Training was a fix for information asymmetry
Think about why corporate training existed at all. Knowledge was locked up. When SAP arrived, the knowledge of how to use it lived with certified consultants. When the PC arrived, it lived in manuals and night classes. The company had to buy access to scarce information and distribute it to employees. Training was a supply chain for knowledge.
Each technology wave since has shortened that supply chain. Books gave way to the internet. The internet gave way to Stack Overflow and YouTube, where the answer to almost any how-to question was free and a search away. With each wave, the training department mattered less and self-directed learners pulled further ahead.
AI is the terminal case. It is the first technology you can ask how to use itself. Stuck on a prompt? Ask the model to improve it. Don’t know what a tool can do? Ask it. The information asymmetry that justified training programs is not just low. For this technology, it is zero.
So when a company designs an elaborate AI curriculum, it is building a supply chain for something that is no longer scarce.
What is actually scarce
If everyone has free access to the same information, why do adoption rates inside a company vary so wildly? Because the bottleneck was never information. It is motivation and agency. The willingness to try, to break your own workflow, to feel stupid for a week while you rebuild how you work.
No curriculum produces that. I have watched people finish an AI workshop and go back to working exactly as before. And I have watched engineers with no training rebuild their entire workflow in a month because they wanted to.
You cannot train agency into people. But you can create the conditions where it surfaces. Which brings me to the only adoption plan I think a company needs.
Publicize the top 5%
In any organization, some small group is already creating real impact with AI. Not the loudest people. Look at output. Find the top 5% and publicize what they did, relentlessly and specifically.
Two rules make this work.
First, publicize the artifact, not just the person. The demo, the workflow, the prompt, the before-and-after numbers. Applause creates envy. Artifacts create copies. The goal is that someone two teams away sees exactly what was done and reproduces it the same afternoon.
Second, let status do the distribution. When the people getting visible recognition are the ones using AI with agency, everyone with ambition recalibrates on their own. No mandate required. The motivated ones will find the information themselves, because the information is free. That was the whole point.
Leadership’s job in this is small and specific: pay for the tools, settle the data and security questions once, remove the procurement friction, and run the spotlight. Clear the path and amplify. Everything else is bottom-up.
A player of yesterday, A player of tomorrow
The uncomfortable half of this: what about the people who, with the tools paid for, the path cleared, and the examples publicized, still don’t move?
Companies have always hired A players. Nobody ever ran a training program to turn C players into A players; they set a bar and hired to it. AI has not changed the strategy. It has changed the test. Agency with AI is now part of what an A player is, and an A player of yesterday might not be an A player of tomorrow.
To be precise about who I mean: not the team in a compliance-heavy function that adopts slowly because it must. Not the person who tried, struggled, and is on their second attempt. I mean the person who, after the asymmetry is gone and the path is clear, chooses not to engage. That is not a training gap. It is a bar, and companies that hold it will outrun the ones running workshops for people who have already decided.
The plan, in full
Buy the tools. Clear the path. Find the top 5%. Publicize their artifacts until copying them is the obvious career move. Hold the bar for everyone else.
That’s it. The information is free now. What you’re selecting for is the will to use it.
