Google keeps shipping Gemini upgrades under one branded name change after another, and most people have no idea which version actually answered their last question. I hope we reach a point where the model name and number (regardless of the AI provider) are much easier to identify. If models are chosen automatically based on the intent and request of the prompt, some barriers to entry may fall.
We see how chatbots like Dots, Muse, or Duet take a request and run with it, without you having to decide whether deep thought is required or whether the request is straightforward. We need this for regular AIs like Codex, ChatGPT, Claude, and Gemini.
I don’t want to have to ask an AI agent which model to use and why. I do that now, and it gets me down a path, but then I sort of lock into that model and might miss newer ones. Yes, being able to select is important, and I do think it is getting more streamlined, but I still feel that, for the everyday consumer, this needs to cause a bit less friction.