You gave everyone the same AI. Why did it make your best people pull further ahead?

You rolled out the same AI to everyone, and you expected it to level the field. Give the strongest tool to all your people, the thinking went, and the gap between your best and the rest should close. Then you watched what actually happened. Your strongest people got dramatically faster. Your struggling people got a little help and a lot of confusion. The gap you meant to close is wider than when you started, and you are not sure why.

Here is why, and it is fixable. A powerful general tool does not lift everyone evenly. It rewards the people who already know how to use it, and those are your most capable people. Handed out and left alone, AI compounds the strengths you already have. But that outcome is a choice you can change, once you see the mechanism.

Does AI close the skills gap or widen it?

It depends entirely on how you deploy it, and the usage data is starting to show the default. Anthropic's Economic Index is its ongoing study of how the Claude model is used. It reported in June 2026 that use is still led by augmentation, where the tool helps a person, but that automation is rising, especially in technical work. Its earlier analysis found that experienced users extract far more from the tools than newcomers do. They know what to ask, when to trust the answer, and how to fold it into real work. Beginners do not, yet.

So the default direction of a general tool, left to self-directed use, is to widen the gap. The fluent get faster; the rest get noise. Read the scope fairly, though, because the picture is not one-sided. In other settings, a tool built for a specific job has lifted beginners the most, by handing them the habits of the experts inside the task itself. Both are true. The difference is not the AI. It is whether the deployment was designed to lift the many or left to reward the few. It is the same lesson behind treating knowledge, not the tool, as the real advantage.

Why does a general tool reward the already-capable?

Because using AI well is itself a skill, and it sits on top of the skills you already have. To get value from a general tool, you need to know what a good question looks like, when an answer is off, and how to turn it into finished work. Your capable people have that judgement, so the tool multiplies it. Your struggling people do not yet, so the same tool hands them fluent output they cannot fully evaluate. The AI did not create the gap. It amplified the one already there.

The lower-altitude question is which AI tool to buy, and how many licences. The higher one is who your deployment is actually going to lift. Buy the best tool in the market, hand it out flat, and you will widen your internal inequality while believing you invested in everyone. That is the same trap as giving people a powerful new tool and never teaching them how to use it well.

So how do you make AI lift the whole team?

You treat fluency as something you build on purpose, not something you hope people pick up. The aim is to give the many the judgement the few already have, so the tool multiplies more of your team. Work it in this order:

1. Build fluency deliberately. Teach people what good use looks like, with real examples from your work, rather than handing over a licence and hoping.
2. Aim tools at specific jobs. A tool shaped around a real task, with the good patterns built in, lifts beginners far more than a blank general assistant does.
3. Spread what your best people know. Capture how your most fluent users work with AI and put it in front of everyone else. That knowledge is the thing that closes the gap.
4. Support the people furthest behind. Give extra help to those who gain least by default, or the tool will quietly leave them further back.
5. Watch who is actually being lifted. Track whether AI is raising the whole team or only the top, and correct the design when it drifts.

What does this make possible?

A team that rises together, rather than splitting into the fluent and the left-behind. When you build fluency on purpose, aim the tools at real work, and spread what your strongest people know, the gains stop pooling at the top. Your capable people still fly. Your developing people climb faster than they could alone. The gap narrows because you designed it to, and the whole team gets more from the same tool. That is a stronger outcome than letting AI amplify only the strengths you already had.

Picture the business a year on. You handed out the same AI, but this time you built the fluency around it, and the results are spread across the team rather than concentrated in a few stars. The people you worried you would lose are keeping up and growing. Fluency turned out to be a thing you could build, not a lottery you had to accept. That is what work looks like after business-as-usual, when a leader decides who the technology is going to lift.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.