I care far less about which AI is the smartest and far more about which one stops making things up, and “half as many mistakes” is the first model pitch in a while that speaks to that. That being said, I have AI update fatigue. I can only imagine what other consumers feel. And while most people will just “take the default” version, I wonder if that is the best thing they can do.
I still go back and ask AI (normally Gemini because it’s Google and Google indexes everything) what the best current model is for my tasks. It gives me the pros and cons of each model, and then I choose from there. So far, that has worked quite well.
The other problem is that if you are midstream with one model and switch to another, it’s almost like having a new employee join your ranks. You have to adapt and learn how to work with them.