The thing that’s broken is capitalism. Most of the shit sorta worked ok 10 years ago (after several decades of pretty shoddy function).
But the tech overlords can’t leave it alone. They must chase “innovation” ie the next big thing. Thus forced upgrades and planned obsolescence and AI and 🐂💩
OpenAI would love for people to believe AI is saturating things, but outside this use case of smarter-autocorrect, there has been no penetration. No one is adding this to their app to do things behind the scenes, it's all theater.
Just a little splash, absolutely no saturation
And TBF, there’s a *lot* of hidden AI in background business services, and growing rapidly. Stuff users never see and may not even have an interface or output long form text. It’s pretty huge.
Yes of course many machine learning methods are used extensively and are useful and effective and can often use less energy than a declarative program!
LLMs are not one of them.
I’m not defending the practice, but LLMs are used in quite a number of areas in backend systems now and continue to grow. Vector embedding w/ semantic matching on its own is pretty huge. I’ve been involved w/ a few large scale LLM projects where no consumer would ever know they were there.
I think the consumer facing AI is mainly what people are talking about when they discuss AI running amuck, tbqh.
SEO, writing, artwork. that sort of thing.
