Does using AI at work mean less time talking to colleagues and more time alone writing documents?

This explores whether AI changes what kind of work knowledge workers do, moving their time away from talking and coordinating with colleagues and toward producing documents on their own. The short answer from the corpus is yes, but the change is a matter of balance and not a retreat from communication. Heavy generative AI users increased their actions in productivity apps (documents, spreadsheets, slides) by 21.2 percent, while their communication actions rose by only 7.1 percent Does generative AI shift knowledge workers away from communication?. Communication still grew. What changed is that solo production grew about three times faster, so the shape of the workday tilts toward the individual at the keyboard.

One reason this might happen is that AI can stand in for some of what a teammate used to provide. In a field experiment with 776 Procter & Gamble professionals, individuals working with AI produced solutions as strong as two-person teams working without it. The AI also pulled people beyond their own specialty, so a commercial person's proposal looked more balanced with technical considerations, and the reverse Can generative AI replace the benefits of having a human teammate?. If you can get a second perspective without scheduling a meeting, some coordination simply never has to take place. A related argument says AI mostly compresses the middle "execute" layer of work, the drafting and producing, while deciding what to do and delivering it to people keep their size or grow Does AI really compress all layers of knowledge work equally?. Put the two together and you get an interesting tension: if execution becomes cheaper and is increasingly done alone, the communication that remains should be concentrated in decisions and handoffs, where it matters most.

The less obvious angle comes from research on AI agents rather than humans. When multiple AI agents work together, they coordinate better by producing and pulling from standardized documents than by chatting back and forth Does structured artifact sharing outperform conversational coordination?. That suggests the shift toward documentation may not mean coordination is lost. The document itself can become the coordination channel. A well-structured spec does some of the work a meeting used to do.

There's a catch, though, and it's where the corpus gets most interesting. Documents work as coordination tools because they're supposed to reflect someone's thinking. One line of argument holds that AI separates the polished outward form of intellectual work from the reasoning that would normally produce it Does AI separate intellectual form from the thinking behind it?. Readers then do interpretive work to recover a meaning that may not be there Does AI generate genuine utterances or just text patterns?. If teams coordinate more through documents just as documents become less reliable signs of thought, communication doesn't disappear. It moves to a new question: how did you make this? Writers who share editors already show this. They strongly preferred seeing when and how their collaborators used AI prompts, because it helped them understand their partner's thinking and check the text Do writers want to see each other's AI prompts in shared editors?.

The corpus gives clear evidence of the rebalancing but less on its long-term effect on team relationships, trust or how organizations learn. Those effects remain open questions. Here is the takeaway you might not have expected: AI may not shrink collaboration so much as change its subject. Teams talk less about producing the work and more about where the work came from.