Popular - subjective. I'll clarify the criteria as things unfold.

The idea is that if you can give image inputs to GPT-4 and get back natural language description of the scene, and you can use the description to do task planning, etc, then integration into a smart glasses which processes things you see in real time, it could be v valuable even in jobs which don't directly require one to work with computers, or aren't easily automated (i think the phrase is blue collar work is the phrase, but I'm not familiar with it).

The integration doesn't have to output text to the smartglasses screen. It could take a sequence of actions, have voice integration, or produce outputs in other modalities or forms. Something like huggingGPT would also work.