Quote:
Originally Posted by ncole1
And after more than 300 posts, you still have yet to address the problem I explained in post #1360.
You can make all the crazy assertions you want, but until you address this issue, no one has any good reason to believe you.
"Ray Kurzweil said so" is not an argument.
|
You might think an exponential trend today might become flat, like CPU speed. Fine, but it is also you that have to show why it would (IMO).
I actually read the post 1360 and it starts out like this:
> No, but you are continuing to ignore the problem I pointed out.
But well, okay, your stand seems to be that other people must prove that, say, hyper intelligence needs X,Y,Z to become and that those are, say, exponential and will likely stay so. Seems fair.
My stance is that computer power for the dollar is on an exponential trend and it will for sure stay on track for at least 10 years (Intel 7nm etc.) and most probably for far longer (Intel: nanotube transistors or Bust, 3Bn$ research to complete 2020 means people are taking this seriously).
Moore's law is, as we all know it, not a law in itself, it was just an observation but over time it has been adopted by the industry more like a planning. An industry that invest enormous sums of money. That makes me think this trend will continue a long time.
We also need the processing power, is is not just a research thing or for glory, the country with the biggest cpu power is better off than another country with less. Again, people/governments are not letting this down in the foreseeable future.
What do we need to simulate a human brain?
Exascale computers and simulation software.
Exascale computers are to be created 2018-2019.
By 2020 (or 2021?) the European Brain project will have simulated a brain(most probably), there is an American project too likewise IIRC.
2023(low estimate of brain cpu power)-2025(high estimate of brain cpu power) a computer able to simulate a human brain will cost 1000$.
When you can actually have a brain (not concious but fully simulated) running in a 1000$ computer, I can't see how AI won't be able to use parts of a brain to become incredibly smart, not having a consciousness but extremely good at whatever we humans are good at like pattern matching, inventing stories, simulating affection or whatever.
And after that I can't see how smart people won't hack together real AI.
Now, I'm no future teller so I don't know when this super AI will come around (maybe 2025? maybe 2030?) and what will happen with it but put it in a robot and it could do any job we humans do so if it is cheaper than human labour, we would not need to work any more. Obviously it could, for starters, create ... robots.
For the nano tech part, I truly believe Freitas, Drexler and al know what they are talking about and it seems reasonable to see APM by 2030, maybe later, maybe sooner, it is a complex matter to measure, so I rely on the authorities in those fields.
Anti-aging tech falls in this category too, De Grey says maybe 2035-2039 well have full blown SENS. This matters, probably, less as we, hopefully, will have attained LEV even today (except for the elderly).
I don't have the time to talk about synergies (Kurzweils accelerating returns) and I have made my stance without it. I think though that advances in one tech will help oters making this be true maybe at an even faster pace.
Gotta go now but feel free to discuss of course!
BTW, I "predicts" a lot, what is your take on the future?