If AI were banned I'd just go back to not being pressured into using something that is a mixed bag in terms of usefulness. Yesterday I used an LLM to try to clean up some tech debt in a codebase. The result is so many code changes that it feels like I should vet what strategies it employed and then once confirmed they are good strategies, make the changes myself in a cleaner and more surgical way. Which wouldn't use a fuckton of electricity like the LLM did. The way the LLM did things edited over 100 files and added tons of comments that won't matter 3 weeks from now. Reviewing all that feels like more work than if I did it myself. I guess I could just blindly trust it if I was an idiot. In the end it may have taught me a couple of things but it's unclear if this is worth it.