But if you don't have a use case then having this sort of broad capability is not actually very useful

This whole article is just repeatedly hitting the nail on the head but this quote is especially accurate for me for so many things. If you don't have something specific you think any new technology or feature will solve, just move on. It's really that simple. I have become an extreme AI skeptic, not least because it has only ever done one useful thing for me, and that was completely by accident (and I'm not even sure if it was right anyway) (if you're interested, I needed monthly local median sunrise/sunset times for an entire year, and it gave me a table that didn't look ridiculous). Everything you read is about how terrible/amusingly terrible it is, and every time I use it it's just not good.

I've had this conversation with clients so many times about various things. Most specifically reporting. Clients always want to add "reporting", but when asked what are the sorts of things they want to report, they don't know. I've even trained clients on their data and installed Superset, but it doesn't get used because they don't know how to use SQL, or they realise they don't actually know what they're looking for (or they just use the 3 dashboards I made for them as an example that don't really provide any business intelligence).

In general, I find the immediate flocking to any new technology frustrating. I've always been like it with language adoption, preferring to be fashionably late to the party, and it has served me pretty well. AI seems to have been a particularly insidious one because everyone is rushing to implement it with no real idea of what they're doing or why. I'll just stay here repeatedly saying "why" like an irritating toddler until I either get a good reason, or the asker backs down.