When people use generative AI at work, does their time tilt from coordinating with colleagues toward working alone?

This explores whether generative AI pulls knowledge workers away from coordinating with each other and toward working alone. The most direct evidence says yes, but the shift is relative, not absolute. Heavy AI users increased their actions in productivity apps (documents, spreadsheets) by 21.2 percent, while their communication actions rose only 7.1 percent Does generative AI shift knowledge workers away from communication?. People didn't stop talking to colleagues. Their solo output grew about three times faster, so the mix of their work tilted toward producing on their own. AI changes the kind of work people do, not only how much of it.

A likely reason shows up in a different experiment. At Procter & Gamble, individuals using AI produced solutions as strong as two-person teams without AI Can generative AI replace the benefits of having a human teammate?. AI also got people to propose more balanced solutions across professional backgrounds, which is something you usually get by pulling a colleague from another department into the conversation. If AI can supply part of what a teammate supplies, some coordination becomes optional. The question is less 'does AI reduce coordination?' than 'which coordination was only there because one person couldn't do the work alone?'

The less comfortable part is that some of that coordination was also how people learned. In interviews with South Korean software engineers, entry-level tasks that used to be handed down to juniors were absorbed into senior-plus-AI workflows Does generative AI prevent juniors from getting entry-level work?. The handoff disappeared, and so did the hands-on struggle through which juniors built expertise. Another study points the same way: AI-assisted workers did better on the task in front of them but showed no improvement when they worked alone afterward Does AI assistance help workers learn lasting skills?. Less coordination can mean less passing-on of skill. At the scale of a whole society, a related argument holds that institutions stay aligned with human interests partly because they depend on people who care how things turn out. Swapping those people out removes a quiet check Does incremental AI replacement erode human influence over society?.

Two caveats keep this from becoming a sweeping claim. First, the shift depends on deep adoption. At Argonne National Laboratory, AI use stayed small and experimental, mostly structured writing, and few staff had built it into their daily work How are national lab staff actually using generative AI?. Where workers actually hand tasks over to AI, it concentrates in information-heavy jobs Where have workers actually delegated tasks to AI?, so any change in how much people coordinate will be uneven across occupations. Second, coordination may be moving rather than disappearing. Research on AI agents finds that a single agent hits organizational limits on complex tasks, and that splitting the work across specialized agents requires explicit structure to coordinate them Do single agents always hit organizational limits?. The coordination problem may come back as a question of how to orchestrate AI systems rather than how to run meetings.

The corpus has one direct measurement of this shift, so read the trend as early and not settled. The more surprising takeaway is that the coordination AI replaces may not be pure overhead. Some of it was how junior people learned and how organizations kept human judgment involved.