I think Karpathy's assessment of 5,000 people in the world with insider AI knowledge goes directly to what is coming and what the majority are not aware of.
> how businesses will profit from AI remains uncertain...And he said that AI model developers, such as OpenAI, need to find customers that can generate returns from putting AI to work, rather than relying on users who are now “wasting money” on tokens as they experiment with AI
I think this is part of the reason why AI CEOs have pushed for a "slow down". It's hard to keep the hype train moving at speed when people aren't seeing real measurable benefits from the product. After years of extremely impressive sounding promises about what AI will do for businesses, the AI these companies actually have to offer can't live up to it. The longer that goes on, the more people are going to start feeling cheated.
An "AI slow down" would give companies an excuse to point to, so they can say "You would have been able to fire all your employees and replace them with our AI by now, but AI was just so good and powerful that the government had to slow us down, so you'll just have to wait, but keep investing because it's totally going to happen once the slow down ends"
More than firing employees, I think employees will quit workplaces that don’t provide enough AI access. It’s hard to go back to coding everything by hand. Expectations have also gone up a lot.
I think the problem is everyone is on this speeding train heading for a cliff. If you slow your train down then you get left behind and investors hate you. But if you keep up you fly into the ravine. CEOs want to slow down but they're all very afraid of moving in a different direction than the herd.
Back 15 yrs ago when information didn't quite makes it way through to even the moderately dumbest person - you could have. Not anymore. These fellas really believed so hard in their scaling thesis and forget that they need to manage the time-horizon and the probabilities along the way.
I think Karpathy's assessment of 5,000 people in the world with insider AI knowledge goes directly to what is coming and what the majority are not aware of.
https://www.businessinsider.com/andrej-karpathy-ai-gap-widen...
> how businesses will profit from AI remains uncertain...And he said that AI model developers, such as OpenAI, need to find customers that can generate returns from putting AI to work, rather than relying on users who are now “wasting money” on tokens as they experiment with AI
I think this is part of the reason why AI CEOs have pushed for a "slow down". It's hard to keep the hype train moving at speed when people aren't seeing real measurable benefits from the product. After years of extremely impressive sounding promises about what AI will do for businesses, the AI these companies actually have to offer can't live up to it. The longer that goes on, the more people are going to start feeling cheated.
An "AI slow down" would give companies an excuse to point to, so they can say "You would have been able to fire all your employees and replace them with our AI by now, but AI was just so good and powerful that the government had to slow us down, so you'll just have to wait, but keep investing because it's totally going to happen once the slow down ends"
More than firing employees, I think employees will quit workplaces that don’t provide enough AI access. It’s hard to go back to coding everything by hand. Expectations have also gone up a lot.
great point, I never thought of that angle
I think the problem is everyone is on this speeding train heading for a cliff. If you slow your train down then you get left behind and investors hate you. But if you keep up you fly into the ravine. CEOs want to slow down but they're all very afraid of moving in a different direction than the herd.
Yeah I think everyone with two brain cells to rub together has figured this out. Which explains why it’s taking so long to take root with investors!
Nobody is buying this stuff.
Back 15 yrs ago when information didn't quite makes it way through to even the moderately dumbest person - you could have. Not anymore. These fellas really believed so hard in their scaling thesis and forget that they need to manage the time-horizon and the probabilities along the way.
Did you read actually the document about the slow down? It has nothing to do at all with what you are saying
[dead]