People talk to people, AI talks to AI: my rule for reviewing AI-generated work

A simple rule for reviewing client briefs and specs: if a person made it, I read it; if AI made it, AI reads it first. This isn't anti-AI. It's attention routing.

The Golden Rule: people talk to people, AI talks to AI. If you did the thinking and shaped the work yourself, I review it directly. If AI generated most of it, my AI reviews it first, then I review the review.

I have a new rule for reviewing client briefs and specs. If a person made it, I read it. If AI made it, AI reads it first.

Some clients send me a well-structured brief that clearly contains a lot of their own thinking. AI may have helped them shape it, organise it, or sharpen it, but the work is theirs. I take the time to read that properly and review it directly.

Other times, it is pretty clear that someone has typed something into ChatGPT, copied the output, and asked me to go through it.

That is fine. Genuinely fine. But in that case, I usually get my AI to review it first. Then I read the review.

My simple rule of thumb

  • People talk to people.
  • AI talks to AI.

Human-made work gets human attention. AI-made work gets AI filtering first.

This is not anti-AI. It is attention routing.

I think the new industry standard should be this: human-made work gets human attention, and AI-made work gets AI filtering first.

Keep reading — it's free

Pop in your email to keep reading and join AI Dispatch, our newsletter of practical advice for leaders scaling AI. Unsubscribe anytime.

Related