Are major AI tools diverging?
It seems to me that Claude, Codex, and Gemini are:
- Better and better, and
- More and more different.
I'd love to give you more precise versions of these, but (i) I doubt it can be measured and (ii) even if it can, I certainly haven't.
This would be great news: models that have different blind spots and can give different effective explanations can be combined in powerful ways. If you've been working with a single tool for a while, it might be a good time to try one or two others and see if your experience matches mine.
P.S.: I think I've read someone saying something similar recently; I had thought it was from Jessie Frazelle, but I can't find any such post from her.