I want the bubble to burst too, but the reality is that LLMs ARE useful, can be very useful in certain tasks, but not at the level everyone was expecting. This bubble won’t burst, but it will deflate a bit, not as much as we’d want. It’s here to stay i’m afraid.
can be very useful in certain tasks, but not at the level everyone was expecting. This bubble won’t burst, but it will deflate a bit
I almost got whiplash reading this.
Yes, you are correct, generative AI is useful for some tasks. Not at the level everyone was expecting is somewhat of an understatement, as, for example, the SpaceX IPO estimated LLM would be worth roughly 200x what even rose-colored projections of the entire LLM market predict. And that’s one company versus the entire market. Being overvalued by 200x does cause bubbles to deflate A BIT.
LLMs will remain, the tech is out there and plenty of open models and weights won’t disappear with the demise of OpenAI and Anthropic.
However, the bubbe will cause severe economic troubles to several companies, not only tech - banks, pension and insurance funds that are buying into the datacenter hype will find themselves with no ROI once said datacenters come online. 1.65 fucking trillion dollars in debt isn’t something that can be handwaved away, and that’s the debt held by, effectively, a dozen companies that are failing to get anywhere near enough revenue, or growth, to justify that colossal spending.
I want the bubble to burst too, but the reality is that LLMs ARE useful, can be very useful in certain tasks, but not at the level everyone was expecting. This bubble won’t burst, but it will deflate a bit, not as much as we’d want. It’s here to stay i’m afraid.
I almost got whiplash reading this.
Yes, you are correct, generative AI is useful for some tasks. Not at the level everyone was expecting is somewhat of an understatement, as, for example, the SpaceX IPO estimated LLM would be worth roughly 200x what even rose-colored projections of the entire LLM market predict. And that’s one company versus the entire market. Being overvalued by 200x does cause bubbles to deflate A BIT.
LLMs will remain, the tech is out there and plenty of open models and weights won’t disappear with the demise of OpenAI and Anthropic.
However, the bubbe will cause severe economic troubles to several companies, not only tech - banks, pension and insurance funds that are buying into the datacenter hype will find themselves with no ROI once said datacenters come online. 1.65 fucking trillion dollars in debt isn’t something that can be handwaved away, and that’s the debt held by, effectively, a dozen companies that are failing to get anywhere near enough revenue, or growth, to justify that colossal spending.
The web didn’t go away when the web hype bubble burst. But it changed a great deal, and a bunch of suckers got to be very broke.
This bubble is much bigger, and will hurt much worse when it goes.
Meta just took at 11% stock price bit for saying “we still think AI is a good idea”. This bubble is right on the precipice.
Maybe it can stay irrational for a bit longer, but it doesn’t look good at all.