Exactly. I know it's easy to automatically froth at the mouth with rage when seeing "AI", and here anything mentioning it gets automatically rejected, but there are genuinely good usecases.

Amazing speech synthesis and recognition is useful for anybody, but especially people with certain disabilities.

Much better translation, spell checking, help with writing. Helping people understand texts that are written in a complicated way (legalese, technical jargon, condensing EULA's, etc)

Infrastructure planning and traffic control.

Grid energy usage and distribution.

Image recognition, useful for anybody for things like searching a photo library for a specific thing, but also for people with visual issues who previously had to rely on awful screen reader software that can't tell you the content of images unless it was properly tagged (as someone with a blind sister who uses computers - rare!)

Spotting fake reviews, a massive issue online. Flagging bot accounts.

The potential for them to take over some jobs and free up people to pursue other things in life.

This technology, if trained ethically, and not used to siphon more data from people, is amazing. It's how megacorps are using it that's the problem.