Skip to main content

Been dealing with a recurring problem that I’ve yet to solve.

The sequence broadly follows the same pattern:

1) We submit work to a client for approval. We’re a PR agency, so usually it’s written work.

2) The client bounces that work off AI for feedback. AI replies with its vanilla logic about angles, directions, etc and then suggests copy edits.

3) The client acts as human proxy for AI and passes the macro and micro suggestions our way, largely unfiltered.

4) We spend a bunch of time explaining why AI is shit at writing and generally weak at strategy.

5) We spend more time editing AI’s garbage prose, make some of the changes and move forward.

It’s starting to grate on my brain.

Why?

I am not sure anyone would use AI to double-check the suggested approach and plan of a car mechanic, radiologist or the chef they were buying dinner from that night. I could be wrong. Is this my ego? Fucken right it is. And maybe I should check myself.

More than that, though, AI’s writing is unbelievably hard to edit. I know that’s, again, likely a me thing, though I have seen other writers I respect say this. Bottom line, it’s time consuming as hell and especially numbing because of the way AI slaps words together (its tendency to over-write everything is especially hard as an editor).

Clients, who start from a good place of wanting to help win the day, end up with a weaker product that takes too long to produce because of the time spent editing it. I cannot overstate this extra time factor. It is one of the great ironies of the AI era for our agency.

We have been in biz for 15 years, so there are plenty of established processes. I think we need a new policy that explains to clients why editing written output with AI is not in their interest.

All this has certainly driven home the reality that AI’s greatest threat to humanity is not that it’s coming to kill us, but that it enables dilettantism at scale.