With the way LLMs work, there’s no chance of them becoming AGI.
And the bubble is already starting to pop. Since Anthropic and OpenAI jacked up the prices, the contracts other AI companies who use them on the back end are unprofitable.
Axios reported that AI is now more expensive than the humans it tried to replace.
I’m kinda looking forward to the collapse of the Ai bubble.
Once upon a time, a lot of telecom bubble startups went under and left behind lots of dark fiber optic cable that others snatched up for far less than the capital cost of construction. But eventually, over subsequent decades, most of it was put into use. I expect these AI data centers will be similar, except a bunch of them are being built in stupid ways like with their own on site electricity generation.
I’m not that excited about what people will do with all of the super cheap bankruptcy auction surplus (formerly AI) data compute capacity. But at least it will be a novel problem.
And the depreciation of hardware is magnitudes faster than fiver optic cables. You can’t slowly snatch up hardware over the course of a decade. In a decade it would be electronic scrap.
With the way LLMs work, there’s no chance of them becoming AGI.
And the bubble is already starting to pop. Since Anthropic and OpenAI jacked up the prices, the contracts other AI companies who use them on the back end are unprofitable.
Axios reported that AI is now more expensive than the humans it tried to replace.
I’m kinda looking forward to the collapse of the Ai bubble.
Once upon a time, a lot of telecom bubble startups went under and left behind lots of dark fiber optic cable that others snatched up for far less than the capital cost of construction. But eventually, over subsequent decades, most of it was put into use. I expect these AI data centers will be similar, except a bunch of them are being built in stupid ways like with their own on site electricity generation.
I’m not that excited about what people will do with all of the super cheap bankruptcy auction surplus (formerly AI) data compute capacity. But at least it will be a novel problem.
except the AI chips and equipment are pratically useless for commercial use.
And the depreciation of hardware is magnitudes faster than fiver optic cables. You can’t slowly snatch up hardware over the course of a decade. In a decade it would be electronic scrap.
LLMs don’t need to become AGI. They only need to convince people in charge that they are.