Why hasn't the Singularity gone mainstream (yet?) (power, advance, smartphone)

[ncole1]

Originally Posted by ncole1

I don't care what people said in the 80's. That's not the point, just a red herring.

Also, no one in the 80's did an analysis like Herb Sutter's article either and came to that conclusion. So your analogy is invalid.

[Josseppie]

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.

Lets revisit this in a year as wearable tech becomes more common.

[ncole1]

Originally Posted by Josseppie

Lets revisit this in a year as wearable tech becomes more common.

Ok, I'll be happy to revisit in a year, after another year of stagnant CPU speeds

[Valmond]

Originally Posted by ncole1

The graph near the bottom of page 25 seems to be cherry-picked to me because it includes a set of performance measures for supercomputers over various time frames, but leaves out desktop and laptop computer processing speeds and thus conveniently ignores the performance measures that have been flat for some time now.

Of course, you could make the argument that the Herb Sutter article "The Free Lunch is Over" cherry picks in the opposite direction - and to some extent I would actually agree with that sentiment.

The point to draw from all of this is that, some performance measures have been increasing exponentially (as for instance those on page 25 of that source), while others did so until around 2003 and then stopped, as shown in the Herb Sutter article (Note, by the way, that ILP is included in that graph, so Valmond's counter to the argument is problematic!).

As long as this is the case, you have no basis for asserting that your parameter of choice will increase exponentially over the next "X" years, unless that specific parameter has been doing so for long enough to tell.

What you fail to understand seems to be that a lot of parts make up computer performance.

Halve speed, double efficiency and you have the same performance. Performance is on an exponential run since over a hundred years.

Will it stop? Will it continue? Only the future will tell but I'm betting it will (barring super volcanoes or something). CPU Speed is only a sub measure, ditched by Intel because today's trends is in energy efficiency. Think smartphones here, wearables, laptops, IoT computer nodes, the list goes on where you want energy efficiency instead of Ghz but you know what? Computer performance haven't trembled in its crazy growth by it so I'd say that it is your claim (cpu speen is stagnating so cpu power isn't augmenting exponentially) that is invalid.

[Josseppie]

Originally Posted by ncole1

Ok, I'll be happy to revisit in a year, after another year of stagnant CPU speeds

I will let other people debate the CPU speeds. The main thing I care about is tech getting smaller and faster as it has since i was a kid in the 1970's and the computers took up entire rooms. This decade is now known as the wearable tech decade and the 2020's will be known as the nanotech decade. I plan on merging with the nanotechnology by 2025, only about 11 years away.

[Valmond]

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?

[Josseppie]

The amount of technological change we will see in the next 20 years is going to be very exciting. Why I call us the luckiest generation.

[Josseppie]

One of the things I talk about is how automation will change the service sector and this decade. When I saw this news story I had to post it as things are changing fast now.

This is from CNET:

Would you like some microchips with that burger? McDonald's Europe strikes another blow against human interaction by installing 7,000 touch-screen computers to take your order and money.

The link: McDonald's hires 7,000 touch-screen cashiers - CNET

[Valmond]

Originally Posted by Josseppie

The amount of technological change we will see in the next 20 years is going to be very exciting. Why I call us the luckiest generation.

Definitely!

Except if you are a technofobiac jokes aside, it's really crazy, I already thought that I was in a lucky generation back in the 80-90s with computers and all but what will come will be mind blowing.

[Valmond]

Originally Posted by Josseppie

One of the things I talk about is how automation will change the service sector and this decade. When I saw this news story I had to post it as things are changing fast now.

This is from CNET:

Would you like some microchips with that burger? McDonald's Europe strikes another blow against human interaction by installing 7,000 touch-screen computers to take your order and money.

The link: McDonald's hires 7,000 touch-screen cashiers - CNET

I don't know if it is the norm in the US but here McDonalds (I don't go there very often though but they have this since maybe 2 years) usually have this kind of computers where you can can order and pay, then you queue anyway (with your ticket).

I was at a restaurant this week end (not a Mc Donalds "restaurant") and when the waitress came to take the order, she had this little handheld computer to type in what we wanted. All I could think was, why am I not doing this with my smartphone, bypassing all that? Especially when you try to hail one for an extra coffee and they are really busy...
It could even be like Ãœber, an application that takes care of the billing and all.