AI tools make you faster right now — but does that speed turn into skill you keep, or vanish once you stop using the tool?

This explores what happens to people's skills once generative AI becomes part of their daily work. The usual framing asks whether AI upskills or deskills workers. The corpus suggests that's the wrong choice to be making. Interviews with Dutch knowledge workers found four separate outcomes: some skills develop, some hold steady, some erode, and some get *revalued*. Revalued means the skill is still there but now matters more or less than it used to How does generative AI actually change worker skills?. The same tool produced all four, depending on how each person used it and which of their tasks changed.

The most surprising finding is the gap between performance and learning. Workers using AI do noticeably better on the task in front of them. When they later do similar tasks without AI, that improvement disappears Does AI assistance help workers learn lasting skills?. That changes how to read the well-known BCG result. Lower-performing consultants gained the most from GPT-4, which narrowed the gap with top performers Does AI help lower-skilled workers catch up to top performers?. But if those gains don't carry over to unassisted work, the gap may be narrowing in output while staying put in ability. The same study also warns that when adoption is optional rather than required, the leveling effect can reverse.

Skills don't change in isolation, because the work itself is changing shape. Heavy AI users shifted toward writing documents on their own and away from communicating and coordinating with others Does generative AI shift knowledge workers away from communication?. At P&G, individuals using AI matched the output of two-person teams working without it, and their solutions drew more evenly on different professional backgrounds Can generative AI replace the benefits of having a human teammate?. Put together, these suggest collaboration skills could quietly weaken. That isn't because anyone loses them. It's because the work that used them gets absorbed by the tool. There's an emotional side as well. After working with AI, people felt more in control of their solo tasks, but they were less motivated and more bored. AI had taken over the engaging parts and left them the routine leftovers Does AI collaboration drain motivation when workers return to solo tasks?.

One useful lens comes from the idea of "articulation work": the effort of explaining what you want. Some of it is useful, because putting a task into words sharpens your thinking about it. Some of it is pure overhead, like re-explaining context the system should already know Can generative AI reduce overhead work that adds no value?. Which skills grow may depend on which kind of articulation a tool demands. Many workplaces haven't reached this point yet. At Argonne National Laboratory, use stayed small and experimental, mostly limited to structured writing tasks How are national lab staff actually using generative AI?.

Zoom out to the labor market and the same "it depends" pattern holds, but with structure. When AI affects only a few of a job's tasks, workers can shift toward the tasks it doesn't touch, which cushions job losses Does concentrated AI exposure enable workers to adapt and reallocate?. Whether AI is designed to automate work or to augment people shapes who gains bargaining power Does automating labor concentrate power more than augmenting it?. A cross-domain review reaches the same conclusion about inequality: the direction is set by deployment choices, not by the technology itself Does generative AI inevitably worsen or reduce inequality?. The takeaway is that a worker's skills after AI arrives say less about the AI and more about how their job was redesigned around it. Being more productive with AI and becoming more skilled are different outcomes, and right now the evidence mostly supports the first.