Give a first-year employee and a 25-year veteran two tasks, and let both use the same AI tool.
The first task is to collate the responses from 100 customer surveys and prepare a dashboard that shows what customers said. The first-year employee has never analyzed survey responses before. With AI, that employee can produce work close to what the veteran produces. The AI tool sorts the comments, counts the themes and builds the charts.
The second task is to write a business plan for an existing company that needs a path to growth, working from a written situation analysis. The AI tool will give both employees a polished plan within minutes, but the two employees will use their plans in very different ways. The experienced employee has a much better chance of judging which growth option fits this company, what its customers will pay for and what its people can deliver, along with the taste to know what is worth keeping in the draft. Those abilities build over years of watching plans succeed and fail. In The Value of Knowing Your Business, I described the same pattern: less experienced entrepreneurs who acted on an AI’s generic advice did worse than those who worked without it.
There is a lesson in these examples. When AI is used for entry-level work, it raises the competence of less experienced people. In a customer support study that Stanford economist Erik Brynjolfsson describes, the AI passed along information that the best call center agents already knew. In this study, the least experienced agents improved the most.¹ That is a real benefit to the business, and it levels the playing field for all employees.
Ethan Mollick, a professor at the Wharton School, describes where the advantage of experience materializes. Generating content used to be slow and expensive, and now it is fast and cheap, so the scarce skill is selection, or knowing “which AI outputs to keep, which to discard, and which to use as raw material for something the AI would never have generated on its own.”² He points to a golf pro who can watch a swing and instantly see the golfer’s mistake.
That points to a test for any AI-assisted work in your company: is the output the finished product, or is it raw material? The survey dashboard is close to finished as it arrives, once someone checks the counts. The business plan is raw material. It needs someone who knows the business well enough to see what is generic, what is missing and what is wrong.
It is crucial to set your expectations for AI-assisted work at a level that suits each employee’s experience. A first-year employee can now deliver a solid survey analysis. Expecting the same employee to deliver a business plan with a veteran’s judgment sets them up to fall short and creates unnecessary risk in the business by putting a weak plan in front of decision-makers. When expectations match experience, all employees are able to perform at their best.
Sources
¹ Erik Brynjolfsson, interviewed by Nicholas Thompson, “What AI Is Really Doing to Jobs Right Now,” The Atlantic, video, May 27, 2026.
² Ethan Mollick, “The Overhang,” One Useful Thing, September 18, 2026. The golf pro example is from the same piece.