The new office etiquette: label AI output and add your own effort

An ever-growing share of debugging notes, documents and code is written by machines, and that has created a new etiquette question for teams, writes Tom Bedor in a short post on his blog. When is it acceptable to forward the output of an AI model to another person to read?

On one hand, a model wired into a company's codebase and documentation often produces genuinely useful output. On the other, engineers now spend a growing part of the day reading AI text, and fatigue sets in: "If I can have a robot say something, so can you." Posting un-digested AI output as though it were your own writing reads as inconsiderate.

The incident that produced the rule

Bedor recalls proposing a design and having a teammate ask an AI to critique it. The teammate forwarded the resulting document with the disclaimer: "I didn't read this, so it might not be entirely accurate." His reaction: if reading it wasn't worth your time, why is it worth mine?

Out of that came the principle he now works by: if you are requesting human attention, demonstrate human effort.

What the rule looks like in practice

AI-generated content still gets shared, but with clear labels marking what a machine wrote and with the sender's own commentary alongside it. For code review requests, Bedor says he always reviews his AI-generated code first before asking a colleague to look at it.

Attention was a scarce resource before AI and is scarcer now. Clear labelling and visible human effort, he argues, are a basic form of consideration for teammates — and keep a touch of humanity in the work. In a footnote, he notes that he wrote the post himself, by hand.