Brockman notes people dislike AI coworkers reaching out for help

Bulletin. Simon Willison, on August 1, reposted a comment by Greg Brockman, president of OpenAI, describing internal reaction at that company to AI agents acting as coworkers.

Brockman wrote that many OpenAI staff connect their ChatGPT instances to Slack. He reported that colleagues consistently dislike it when a coworker’s ChatGPT reaches out to them directly asking for help with a task, even in cases where they would be glad to do that same work if the human coworker had asked them personally.

Brockman characterized this reaction as evidence of how much people value human relationships and mutual help. He said the pattern shows a preference for AI that returns time to people, or that enhances time spent together, rather than AI that inserts itself as a layer separating colleagues from one another. The quoted text, drawn from a post on the platform X, ended there in the material available to this desk.

Willison’s post consisted of the quotation itself, presented without additional commentary of his own, under the heading “Quoting Greg Brockman.”

No further detail on the scale of the Slack integration, the frequency of the complained-about behavior, or any resulting change in OpenAI’s internal practice was included in the sourced material.