AI Looks Like a Bubble
every.to
every.to
Yes it's wild that OpenAI raised money at 1,000X earnings but the tech giants keep mountains of money on hand at all times for just these moments.
Right now it's a "jump ball" in just about all tech. LLM threaten all the big incumbents because AI is racing "across the chasm" much faster than most people expected. AI is a new baseline expectation for customers and Google doesn't appear ready for that. It's unlikely other big incumbents are ready either.
For many years we've been working on machine learning to enhance the productivity of our client's employees, now it's almost possible to just replace those employees. AI will do to white collar work what the assembly line did to blue collar work. The ethical implications of that should be considered by people smarter than me.
Now to be clear I'm not crazy! These architectures are not going to become self aware or anything, and this article while thoughtful spends a lot of words picking on C3... whose pitch as best I can tell is just short of a manic fever dream promising an all knowing general intelligence. That's a dead end, and yes like any gold rush a lot of money will be lost but a lot of money will be made as well.
I for one am glad to see the big boys scared, and I'm thankful I get to have another one of these disruptions happen during my career.
Once more unto the breach!
Forgive my skepticism, but AI hype today looks remarkably similar to the NFT hype of yesterday. IMO there's certainly value in AI-assisted workflows, and maybe the hype around AI is doing legitimate work here a disservice, but I don't see any lasting, meaningful impact being had as a result of this wave.
If so, then it is going to kill AI trust for many applications and uses.
The AI is already producing things of production value. People use the art that comes out of it. There's a great demand for mediocre art -- product logos, book illustrations, making web pages feel less empty. It's not great art, but it's the kind of thing you'd have to pay a fair bit of money for.
It's less clear to me that the language models are worth much. Unlike art, there's no great clamor for more mediocre text. At best they're a nod in the direction of something more interesting, in a way that NFTs never did. They may yet burn out without producing anything of value, but it feels like there could be something there.
So you could be right that there's no lasting, meaningful impact. But I don't think I'd compare it to NFTs, which existed solely to push the cryptocurrency bubble a bit further. (Cryptocurrency itself did seem to have some promise, and it yet may. It seems to flounder not on a fundamental uselessness, but more on not providing sufficient value to replace the inertia existing currencies. It was attractive only to illegal transactions, and even that was swamped by pure speculation.)
Eventually the bubble will collapse. Some will win. Most will lose. The value that the winners will win comes from the "gold" actually being mined -- though it's always unpredictable where the gold actually is. Some will lose for backing a reasonable candidate that just didn't survive the "jump ball". Others will win despite backing a mediocre candidate just because it happened to be the one who survived.
I do agree that there is some gold in them thar hills. I could be wrong, but it feels like at least some real value exists. It could be a lot, or it could be a little. I'd love to see it be a lot... but I don't know who will be the one to come up with it. It's often not the early leaders, and it's often not where the sensible bet seems to be.