Borrowed judgement

Somewhere in the last two years most owners stopped checking. Not deliberately. AI arrived, it was obviously useful, and three assumptions went in with it: that it was safe, that it could only make things better, and that it would fill whatever gaps the business had.

It does not fill gaps. It hides them.

That is the argument of the first video, and it is the idea the whole of Qualter+Co is built on. AI has no opinions. It has volume. Whatever thinking you hand it, you get more of, faster, and beautifully formatted. If the thinking was sound, that is a gift. If it was half-formed, you now have a confident, polished, published version of a muddle, and no easy way to tell the difference from the outside.

I call that borrowed judgement. Someone else’s thinking, or your own from three years ago, scaled by a machine. Its opposite is owned judgement: you worked it out, you can show your working, and you would still back it if the model disagreed with you. The video is about how to tell which one you are running on.

One claim in it is worth putting in writing, because it is the one I expect to be argued with. There will be casualties: businesses that adopted fast, scaled a gap, and found out late. Nobody is counting them, because the people selling AI have no reason to and the people who bought it will not admit it. If someone does count, and the number is small, I was wrong. My bet is that nobody counts.

If you want to know where your own gaps are, the diagnostic is free. Twenty-eight statements, fifteen minutes, and a result written by me rather than generated.