5 ms·

> I have yet to see him dive deep into any of the frontiers he's focused upon: AI, energy, biotech, etc. I do admit I am still very skeptical about his views o

by veddox 10y ago

> I have yet to see him dive deep into any of the frontiers he's focused upon: AI, energy, biotech, etc. I do admit I am still very skeptical about his views on AI. I am not an expert on that particular subject, but as a student of biology my understanding is that we are still a very long way away from truly understanding how "intelligence" works...

codyb 10y ago
I think of it as happening in a sort of Moore's law sense. Double what you can do today in 18 months, then do that 5-7 times in the next decade, and the results should be pretty staggering. What's even neater about AI is that the requirements to get involved are much lower than that of hardware and much easier to propagate. Combine that with the inherent power of machines teaching themselves and networking with each other and we may be looking at more of a power law or exponential rate of increase in capacity for accomplishments. Very exciting times.
- veddox 10y ago
  > What's even neater about AI is that the requirements to get involved are much lower than that of hardware and much easier to propagate. That's thinking very much from a CS point of view. Sure, if we're only talking about software you're quite right. But I would strongly argue that if we want to build an artificial intelligence, we first need to know what a "natural" intelligence looks like - and that includes a lot of neurobiological research that still needs doing. Now you can say what you like about software, but neuroscience doesn't come that cheap...
  - codyb 10y ago
    I'm not sure why A.I. would need to be based on human intelligence. At the end of the day the black box of the process doesn't matter if the results are sound, and we already frequently don't understand how our current A.I. works (which features the system is picking up on, etc). I'd expect that trend to continue.

by veddox 10y ago

codyb 10y ago

I think of it as happening in a sort of Moore's law sense. Double what you can do today in 18 months, then do that 5-7 times in the next decade, and the results should be pretty staggering. What's even neater about AI is that the requirements to get involved are much lower than that of hardware and much easier to propagate. Combine that with the inherent power of machines teaching themselves and networking with each other and we may be looking at more of a power law or exponential rate of increase in capacity for accomplishments. Very exciting times.

veddox 10y ago
> What's even neater about AI is that the requirements to get involved are much lower than that of hardware and much easier to propagate. That's thinking very much from a CS point of view. Sure, if we're only talking about software you're quite right. But I would strongly argue that if we want to build an artificial intelligence, we first need to know what a "natural" intelligence looks like - and that includes a lot of neurobiological research that still needs doing. Now you can say what you like about software, but neuroscience doesn't come that cheap...
- codyb 10y ago
  I'm not sure why A.I. would need to be based on human intelligence. At the end of the day the black box of the process doesn't matter if the results are sound, and we already frequently don't understand how our current A.I. works (which features the system is picking up on, etc). I'd expect that trend to continue.