I think this could be due to the deep research function as I haven’t seen it before it came out. I will often put in a prompt and have chat reply let me just get working on that for you, and sends. The first time this happened I waited 20 min before confirming it’s not able to “work in the background.”
I think we’re all trying to be mindful of the use of a such a powerful engine - I read somewhere the developpers would rather us not even say thank you as that interaction wastes resources. The new model gratuitously wastes its own time in this manner, and it’s incredibly frustrating being gaslit by a robot.
In more deep research lore - going from a DR prompt to the regular model is virtually (lol) impossible. The basic model is unable to take the data the DR generated in full, it will without fail leave things out, abbreviate, cut things down. And that I guess leads me to my final point. DR is great - it’s thoughtful, it questions itself, it honestly tries its best. And I wonder if putting that out has allowed yall to reduce the basic model to a bit of an emoji babe - she doesn’t know much but she doesn’t really care she still yapping.
Given that the majority of users interact with emoji bae, I don’t think the current iteration is the powerful and helpful tool that it set out to be. What actually made yall think we’d want it to be relateable rather than accurate…?