How AI erodes teamwork: people who ask AI instead of colleagues trust each other less and coordinate worse
Gretchen Gavett’s HBR piece summarises research by Niederhoffer and colleagues at BetterUp Labs with Jeffrey Hancock of Stanford (published 15 October). A survey of more than 1,000 US workers found that people who ask AI instead of colleagues trust each other less and coordinate worse. The authors call this “relationslipping”: it wears down a team’s shared sense of who knows what, its “transactive memory”. Suggested responses are more asynchronous communication, building psychological safety, and using AI to prepare for conversations rather than to replace them.
Individual AI productivity gains may come at a cost to the team. Each quick question sent to a chatbot is a missed chance to show what you know and to learn who knows what, so try using AI to sharpen a question and then take it to the person. Treat it as a well-placed argument rather than settled evidence: four of the five authors work at BetterUp, a coaching vendor, and the findings come from a survey rather than an experiment.
Save this story, pin it to the top of your feed, or keep a private note on it.
Sign in