@neil I think much of it is - but not all; people are finding genuine good uses for some of the LLMs and other AI; only in a few of the places being tried - but still, those are useful. So yep, it'll implode, and the lessons of the 2000 dot.com bubble is that there will be a lot of bankrupt data centre companies with part built setups. For legal, I'd bet it's pretty good for taking a human written document and suggesting additions for a human to check.

@neil The way you wrote it, makes you sound like a lawyer.

Lawl

@neil Stuff like ChatGPT can I think really only go two ways
- Either it continues to be mindbogglingly expensive and unreliable in which case there's no viable business model
- It or a subset of it can be done cheaply enough that it can be done locally, in which case there is no business model for cloud based generic "AI"

As with .com though most of the "AI" market (except by stock value) is boring, sound non hyped stuff like analysing images or voice recognition. Stuff nobody even notices.

@neil To me, it has a “Second Life” vibe to it. It will probably hang around for a while and several serious people will try it out, use it for a while and then discard it when they realise that the “maintenance cost” of using the output professionally is higher than the benefit they get from it. But that may still take a little while.

@neil I view it like the dotcom bubble of the late 90s: lots of money being ploughed in and 95% of those companies will not have a viable product or will be gone in 5 years.

There will be some companies that survive with some products that are useful, but it will be hard to predict from now.

There will be some cultural change, but it won't totally rewrite our society.