

True, but I got two problems with that thought chain:
- I don’t want any outdated dependencies within my network. There might be a critical bug in them and if I back up the images, I keep those bugs with me. That seems pretty silly.
- If an application breaks because you updated dependencies, you either have to upgrade the application aswell or got some abandonware on your hands, in which case it’s probably time to find a new one.




I wouldn’t say there is “no” use case.
For example, a friend of mine is landscape gardener and he regularly uses AI to basically help customers understand what he plans and how it would look. I made him a small webapp where he can simply snap a pic and writes a prompt like: “To this garden, add a small, round base filled with pebbles in the middle, add a bench and two metal poles left and right”. AI generates the pic, he rewrites the prompt a few times and once he’s satisfied, can show it to the customer who then exactly knows how it looks. He used to do that with photoshop himself which took him a lot longer and customers were more often unhappy with the outcome because the picture didn’t show it as clearly as the AI generated pictures do, so he had to do some adjustments, which obviously eats into the profits.
So yeah, there is actually a good amount of use cases. However, none of those use cases requires literal trillions being pumped into the AI industry.