Is AI a Bubble
It really does not make sense spending trillions on compute. Is this suppose to be a hype for marketing. The way I see it, the price of AI is going down and getting cheaper every quarter. What can we expect in next couple of years. Also why is openclaw.ai landing page down?
7 comments
The product is real (even if it's not as perfect as the companies claim, it has uses), but the imagined future income that the companies are spending already are out of this world. OpenAI needs literally a trillion dollar in the next 3-4 years or so - and they are just one company.
openAI closed down Sora too? How are they spending that much on compute. Anthropic seems realistic.
I picture it as a bubble inside real foundation. The internet was a bubble too but it popping didn't mean that it went away. It meant that the hype was running the treadmill faster than societies cadence.
AI is the way of the future. The roll-out and adoption will be slower than what the "market" has banked on.
AI is the way of the future. The roll-out and adoption will be slower than what the "market" has banked on.
what do think AI will be in 5 years. currently CODING is solved.
CODING feels like the first and easiest step it could take after solving natural language so it is taking the path of least resistance to what it can understand at the "neuro" level.
What's next for it to understand? probably spacial/geometric and mathematical relationships. It's not so good at those still. That will likely take CAD and 3D modelling out pretty quickly etc.
The other thing we are seeing is instead of training it to understand these concepts, just build things towards what it already knows like code and natural language. That is why so much software now wants to have all their stuff be able to be build using code.
What's next for it to understand? probably spacial/geometric and mathematical relationships. It's not so good at those still. That will likely take CAD and 3D modelling out pretty quickly etc.
The other thing we are seeing is instead of training it to understand these concepts, just build things towards what it already knows like code and natural language. That is why so much software now wants to have all their stuff be able to be build using code.
It probably is, since the benefits to common people are still to be known, but a clear value seems to be visible from recent robotic projects where the robots are automating and making things cheaper for people. Though the problems with monopolies may exponent themselves.
there is obviously productivity increase with AI but it might have a ceiling. AI leader bluffing spending trillions is a joke. cost of ai gets cheaper, compute spend getting higher is just confusing. the gap is really huge. it seems just selling hype