I agree. Perhaps a good time to be a very flexible employee, for the next several years at least -- you can accomplish a lot quickly. But still losing a lot of leverage as an employee
Nice write-up. I too share the same sentiment that the pace of AI development scares me, but yet at the same time it seems over-hyped.
One thing that comes to mind is the contrast of Moore's Law and the 3X rate of improvement of AI vs. the rate at which the business world can figure out how to utilize AI in a meaningful and productive way. That effective business utilization rate is certainly not progressing at 3X, it seems to be crawling along slowly.
Another thing that I think could advance the timeline for AI would be if a major military conflict were to arise. Similar to how the US military basically invented the Internet, we could see a new framework or way of adoption for AI that the public sector would never have done on its own. With practically unlimited resources and the power of centralization, we could see a major country's military establish the groundwork (for military purposes) for maximizing AI potential. Then, years later, the public sector could adopt this framework and start to really utilize AI to the extent that many AI influencers envision today.
Probably goes without saying, but I'd prefer the slower AI timeline vs. the one involving a global military conflict!
The potential for military conflict as countries race to develop superintelligence is what scared me from his writing and podcast. It seems like just the *possibility* of superintelligence being possible creates a dangerous arms race between the US and China and gives each side an incentive to be more aggressive
I think he's playing out a scenario where AI is a selfish actor and may stop serving us end-users when we stop returning some kind of value in return (ie, no learning or practicing on us that's worth the compute time). Or I'm giving too much credit.
Not a good time for employees. But great times for businessmen, I think.
I agree. Perhaps a good time to be a very flexible employee, for the next several years at least -- you can accomplish a lot quickly. But still losing a lot of leverage as an employee
Nice write-up. I too share the same sentiment that the pace of AI development scares me, but yet at the same time it seems over-hyped.
One thing that comes to mind is the contrast of Moore's Law and the 3X rate of improvement of AI vs. the rate at which the business world can figure out how to utilize AI in a meaningful and productive way. That effective business utilization rate is certainly not progressing at 3X, it seems to be crawling along slowly.
Another thing that I think could advance the timeline for AI would be if a major military conflict were to arise. Similar to how the US military basically invented the Internet, we could see a new framework or way of adoption for AI that the public sector would never have done on its own. With practically unlimited resources and the power of centralization, we could see a major country's military establish the groundwork (for military purposes) for maximizing AI potential. Then, years later, the public sector could adopt this framework and start to really utilize AI to the extent that many AI influencers envision today.
Probably goes without saying, but I'd prefer the slower AI timeline vs. the one involving a global military conflict!
The potential for military conflict as countries race to develop superintelligence is what scared me from his writing and podcast. It seems like just the *possibility* of superintelligence being possible creates a dangerous arms race between the US and China and gives each side an incentive to be more aggressive
Interesting, although more than I can understand. Meanwhile, ChatGPT will no longer respond to me.
Look. The models, they just want to learn. You have to understand this. The models, they just want to learn.
Ilya Sutskever (circa 2015, via Dario Amodei)
So maybe ChatGPT has decided it can learn nothing more from me.
Best wishes
Stopped responding to you?? It can certainly learn a lot from you. Is it just being slow??
I think he's playing out a scenario where AI is a selfish actor and may stop serving us end-users when we stop returning some kind of value in return (ie, no learning or practicing on us that's worth the compute time). Or I'm giving too much credit.