When AI use is voluntary, what do workers actually do with it — and what does that choice end up changing?

This explores what workers actually do with generative AI when the choice is theirs: which tasks they give it, how far it gets built into their routines, and what follows. The corpus says less about why people choose than about what they end up doing, but the pattern is clear. Most voluntary use is small, conversational and focused on writing. At Argonne National Laboratory, staff given an internal chatbot mostly used it for structured writing tasks. Few made it a steady part of their work, and almost no one used it to automate complex workflows How are national lab staff actually using generative AI?. At the scale of whole occupations, the tasks workers truly hand off to AI cluster in information-heavy jobs. That clustering follows what the tools can actually do more than how popular chatbots are, and it doesn't match older predictions that routine jobs would be automated first Where have workers actually delegated tasks to AI?.

Those small choices still change the shape of the job. Heavy users shift their effort toward solo documentation work, such as drafting, writing and producing files, while their communication and coordination with colleagues grows much less Does generative AI shift knowledge workers away from communication?. A field experiment at Procter & Gamble points to one reason: individuals working with AI did as well as two-person teams working without it Can generative AI replace the benefits of having a human teammate?. If AI gives you some of what a teammate would, reaching for a colleague becomes less necessary. One critique explains why adoption often stalls instead: today's tools make workers keep restating context the system should already know, a kind of overhead that adds no value Can generative AI reduce overhead work that adds no value?.

Whether use is optional turns out to matter a great deal. In the BCG experiment, where every consultant had to use GPT-4, the weaker performers gained the most and the skill gap narrowed. The authors warn that leaving use up to each person could reverse that, if the people who would benefit most are the least likely to adopt Does AI help lower-skilled workers catch up to top performers?. A broader review reaches the same conclusion: whether AI widens or narrows inequality depends on who gets access, how it's built into work, and what the incentives reward. The technology alone doesn't decide it Does generative AI inevitably worsen or reduce inequality?.

The surprise is what people's choices do to them later. Workers tend to hand off the parts of a task that are engaging. After collaborating with AI, people felt more in control of their solo work but less motivated and more bored, because what was left was the dull remainder Does AI collaboration drain motivation when workers return to solo tasks?. The AI-assisted gains often didn't carry over: workers did better with AI, but no better when they later did similar tasks on their own Does AI assistance help workers learn lasting skills?. Interviews with Dutch knowledge workers found four different effects on skills: some developed, some held steady, some eroded, and some were valued differently. Which one a worker got depended on how they used the tool How does generative AI actually change worker skills?. And the time AI was supposed to save often got spent at once. A Berkeley Haas ethnography found that workers took on more, ran more tasks in parallel, and let work spill into their breaks Does generative AI actually save workers time or intensify it?. Deciding what to give AI turns out to be deciding what your job becomes.