AI can make you faster right now, but does it actually make you better once the AI is gone?

This explores whether the boost people get from working with generative AI lasts once the AI is gone, or whether it only holds while the tool is in hand. The short answer from the collection is: mostly the latter, with one important exception. Wu et al. found that workers using generative AI did substantially better on content tasks. When they later did similar tasks alone, their performance showed no improvement. The capability stayed with the tool and didn't transfer to the person Does AI assistance help workers learn lasting skills?. The immediate gains are real and can be large. At Procter & Gamble, individuals working with AI produced solutions as strong as two-person teams working without it Can generative AI replace the benefits of having a human teammate?. The question is what those gains leave behind.

The less obvious part is that the cost may be more than skill. It can also be motivation. Four experiments with over 3,500 people found that after working with AI, people felt more in control when they went back to solo work, but they also felt less intrinsically motivated and more bored Does AI collaboration drain motivation when workers return to solo tasks?. The explanation is that the AI had taken over the engaging parts of the task, leaving people with the dull leftover work. So the problem may not only be 'I didn't learn how to do this.' It may also be 'I no longer want to do the part that's left.' An EEG study adds a neurological angle: over four months, people who relied on LLMs showed the weakest brain connectivity, the poorest memory retention, and trouble recalling work they had just produced Does AI assistance weaken our brain's ability to think independently?. The authors call this 'cognitive debt': the effort you skip now is borrowed against your later ability to think independently.

The story is not uniformly bleak, and the exception is worth knowing. In a randomized experiment with 1,174 adults, AI closed about three-quarters of the performance gap between people with more and less formal education. Lower-education participants kept part of that gain even after the AI was removed Can AI narrow the education performance gap?. This suggests that how much transfers may depend on where you start. People who are missing a framework may actually absorb one from the AI, while people who already have the skill mostly hand it off. Interviews with Dutch knowledge workers support this. They point to four different outcomes, not a simple upskilling-versus-deskilling split: some skills develop, some are maintained, some erode, and some get revalued. Which one happens depends on how each person uses the tool and which parts of their job change How does generative AI actually change worker skills?.

There is also a twist in what 'sustained work' even means once AI arrives. A Berkeley Haas ethnography found that AI didn't free up time. It made work more intense: people took on broader scope, lost natural stopping points and juggled more parallel threads Does generative AI actually save workers time or intensify it?. Heavy users also shifted toward solo documentation and away from communication and coordination Does generative AI shift knowledge workers away from communication?. So the question may be shifting under our feet. Working without the AI becomes rarer, and the work that remains for humans is a different mix of tasks.

The takeaway is that 'AI makes you better at the task' and 'AI makes you better' are separate claims, and the evidence mostly supports only the first. The useful design question is not whether to use AI. It is how to use it in ways that build understanding instead of bypassing it, especially for people who already have the skill and stand to lose the most from handing it off.