Human brain has more switches than all computers on Earth

CNET ^ | November 17, 2010 12:03 PM PST | Elizabeth Armstrong Moore

[DWar]

Among the wisest words ever spoken by a mortal.

I know that You can do everything, And that no purpose of Yours can be withheld from You.
You asked, ‘Who is this who hides counsel without knowledge?’
Therefore I have uttered what I did not understand, Things too wonderful for me, which I did not know.
I have heard of You by the hearing of the ear, But now my eye sees You.
Therefore I abhor myself, And repent in dust and ashes.”

[dynoman]

.

[Jeff Winston]

At this point, human beings are generally much better at some things than computers, and computers are much better than human beings at some other things.

Compared to human beings, due largely to the complexity mentioned above, computers are extremely stupid. But their limited brain power operates at an extraordinarily fast speed.

It turns out that you can (to some degree at least) simulate a very smart (but slow) brain using a really stupid (but enormously fast) machine.

There were some people who kind of panicked over the growing "intelligence" of computers as far back as the 1960s and 70s. These people were many, MANY decades early. By their panic (to whatever degree it was panic), they showed that they underestimated the intelligence of human beings (easy to do based on the results of some of our elections) and overestimated the intelligence of computers.

However, the simulations are gradually always getting better.

But it's not the fact that they're getting better that's the key issue. It's the fact that so far that improvement has consistently been exponential in nature, even through previously unanticipated shifts in technology.

Here's what an exponential curve looks like. This is what is happening to the intelligence of computers:

Assuming the curve holds, in 10 years we're likely to have computers 32 times as powerful as our most powerful computers today. In 20 years, 1000 times as powerful; in 30 years 32,000 times as powerful, and in just 40 years, computers may be one million times as powerful as today's machines.

So imagine the most powerful supercomputer in the world - something run by the US Department of Defense. In 40 years, whatever they have may be a million times smarter than what they have today (assuming, again, that Moore's Law holds, which it has so far).

Will the law hold? I don't know. It appears that people hold different views on this. But if it holds for much beyond 20 years from now, it looks like we'll be in for a heck of a ride.

[Jeff Winston]

I thought that Moore’s law was based on semiconductors and saturates at a certain level because semi conductive materials become fully conductive when they get too small. I suppose a change in transistor material could keep things going, giving us 32 years before the machines take over.

We've already seen some significant shifts in technology along the way. Actually, if you read stuff by Ray Kurzweil (I pulled up an old article by him a little while ago looking for more info on this stuff), he points out that integrated circuits are our fifth major computing technology, and the law has held through all of them (previous computing technologies, in order, were: electromechanical, relay, vaccuum tube, and transistor).

And at present there are at least two new computing technologies that may give us significantly more power than our current tech: optical computing, and quantum computing. But it seems that it's often not until you come to the end of one technology that its replacement begins to become workable.

I would say it's been kind of like driving through the mountains. If you look ahead, the road appears to end. Once you get to that point, you can see further and realize that you can keep going, that there's a way around the curve.

[fortheDeclaration]

And ofcourse it just evolved!

[RightOnTheBorder]

Back in undergrad I remember the chemical engineers looking into chemical chains that changed state to function as transistors. I too believe that Moore’s law will hold. Then the next question will be if brute force computing power will operate in the same manner as a human brain once the number of “switches” is matched or exceeded.

[mamelukesabre]

The “hell of a ride” part is just beginning...I believe...with the invention of the kinect pod thingy microsoft just started selling for their xbox360. That little gizmo is going to morph into intelligent autonomous robots in every home. I think that kinect thing is akin to henry ford’s first prototype horseless carriage.

Maybe not the next version of windows, but the version after that is going to be an on-screen intelligent cyber-being which will be seeing us via a kinect like-device.

[SunkenCiv]

Thanks Ernest_at_the_Beach and dayglored! I just with people would turn the switches off when they leave the room, geez.

[Weirdad]

Nice thread, thanks!

[Revolting cat!]

Not my exes brain, that’s fer shore!

[plinyelder]

MALE BRAIN

[Jeff Winston]

Then the next question will be if brute force computing power will operate in the same manner as a human brain once the number of “switches” is matched or exceeded.

It may not. It may be more like an "alien" intelligence.

I do recall reading though, somewhere during the past couple of years, that some new approaches to AI involve a "learning" approach similar to what children go through. So the first "human-level" AI might actually turn out to be quite like a very good simulation of the development of a human child.

All of which can get kind of disturbing, in a hurry, from a number of points of view.

Disturbing from an ethical point of view, for example.

Disturbing also from the point of view that children aren't always the most stable or civilized of beings until they reach maturity. However, I would think that early "human-level" AIs would be contained in terms of their ability to harm others. And for subsequent generations, you could have "older," "more mature" AIs trained, essentially, in parenting the first of a new generation of AI.

The thing is, the first "human-level" AI is going to be very difficult to produce.

And the next billion copies will be extremely easy.

[Jeff Winston]

Actually, I should say “far easier.” You will, after all, require some very sophisticated hardware for those next billion copies.

[Logic n' Reason]

So to be a good liberal, you have to turn all your switches off

Really really "good" liberals (the unthinking kind...the vast majority) are born without switches...I think it's called "brain-dead"!

They're "green" from the gitgo!

[PATRIOT1876]

A human brain would seem to be a bit more complex than a one-celled organism.

“Many famous evolutionists have calculated the odds of a cell or even just the proteins in a cell randomly assembling. These odds (again calculated by evolutionists themselves) so discredit the theory that they typically are not mentioned in discussions of the topic. The famous atheistic astronomer Sir Frederick Hoyle calculated the odds of even just the proteins of an amoeba arising by chance at one in 1040,000, i.e., one followed by 40,000 zeroes (Hoyle and Wickramasinghe, 1981, p. 130). Harold Morowitz, former professor of Molecular Biophysics and Biochemistry at Yale University, calculated the odds that a simple, single-celled organism might randomly assemble itself from pre-existing building blocks as one in 10 to the 100,000,000,000th power, i.e., one followed by 100 billion zeroes (Morowitz, 1968, p. 98). Carl Sagan and other famous evolutionists (including Nobel Laureate Francis Crick, the co-discoverer of DNA) have come to similar conclusions (Sagan, et al., 1973, pp. 45-46).”
(Had to make slight aleration to original article because “10 to the 100,000,000,000th power” did not show properly)
~http://www.apologeticspress.org/articles/3511

God is awesome!

[Ernest_at_the_Beach]

Absolutely!