Why does using AI at work make people write more solo documents and talk to teammates less?

This explores why AI use pushes knowledge workers toward solo writing and document work rather than team communication. Start with the data, and with what it doesn't show. In a large study, heavy generative AI users increased their actions in productivity apps by 21.2% but their communication actions by only 7.1% Does generative AI shift knowledge workers away from communication?. That records a rebalancing. It does not explain it. The corpus has no study that tests the cause directly, but several notes from other fields point to the same explanation.

The clearest explanation is that AI is very good at speeding up one layer of work. Narayanan and Kapoor split knowledge work into deciding what to do, doing it, and delivering it to others. They argue AI compresses mainly the middle 'doing' layer, while deciding and delivering stay the same or grow Does AI really compress all layers of knowledge work equally?. Drafting a memo, a report, or a spec is exactly that kind of work. When the cost of producing a document collapses, people produce more of them. Studies of where workers have actually handed tasks to AI find the same thing: delegation clusters in information-heavy work, and it follows what the tools can do well rather than how often people chat with them Where have workers actually delegated tasks to AI?. Coordination doesn't get the same boost, because no model can sit in your meeting or settle a disagreement between two departments for you. Evans makes a related point: easier tools don't fix the organizational decisions that cut across teams Does easier tool-building actually solve enterprise adoption problems?.

Science shows the same pattern at larger scale, and that's where it gets interesting. Researchers who use AI publish about 3× more papers and get 4.8× more citations. Yet across science as a whole, the range of topics studied shrinks by 4.63% and collaboration between researchers drops by 22% Does AI help individual scientists while narrowing scientific focus?. If office work follows the same path, more solo documents could be the visible sign of something less visible: individuals producing more while teams talk less. Part of the reason may be that the AI starts doing some of what a colleague used to do. Research on what makes an AI feel like a colleague rather than a chatbot points to memory that persists, reusable procedures, and seeing tasks through to the end What makes an AI system feel like a colleague rather than a chatbot?. A worker with that kind of assistant has less reason to ping a teammate.

Documents may also matter more now because AI depends on written context. Unlike conventional software, an AI's working context (the prompt, the chat history, the material it retrieves) keeps shifting and disappears between sessions How does AI context differ from conventional software context?. Writing things down is how you give the AI something stable to work from, so some of this 'documentation' may really be feeding the machine. There is also a quality risk. AI separates the polished look of a document from the thinking behind it Does AI separate intellectual form from the thinking behind it?, so more documents doesn't guarantee more understanding. People do feel the work is more their own when they steer the text closely Does user control over AI text shape feelings of ownership?, which suggests that how hands-on someone is decides whether the extra documents carry real thought.