Humanoid robot learns how to run

I have never read anything on the subject, although the interest is there, but I think the brain, to a certain extent, works a LOT with maths, it's just very abstract to us. Geometry especially, i can see sometimes that my brain develops "patterns" with numbers and geometry, to understand things that have nothing to do with maths and geometry themselves.
Logic and maths seem to be connected very tightly, at least in my experience.

 

there`s 2 reasons why man kind wont "see" a true AI in it`s life time.
a) we gonna blow ourself up before that.
b) when we`ll build an AI that will be inteligent enough it will blow us up.
in the end, "Artificial intelligence is no match for natural stupidity"

 

Ok then!

No, really, as long as we remember to put a line for Love Donuts in the AI code, robots will only ever kill in situations where they are refused donuts. Which they can't eat anyway but that's not the issue here.

 

c) after it blows us up, it'll blow itself up

EDIT:
so many blowjobs...

 

Speaking of artificial intelligence, maybe someone wants to take a look at this:

SHRDLU Resurrection

...And tell me how to get the Java version to work under winxp.

 

Based upon old estimations from the 70s/80s, which assumed neurons were only capable of one calculation per burst, the brain in a vegetative state was assumed to have the equivalent processing power of approximately 100 TeraFLOPS (roughly 100 trillion floating point calculations per second). This is based on factoring the capability of the brain's 100 billion neurons, each with over 1,000 connections to other neurons, with each connection capable of performing about 200 calculations per second.

As a comparison: http://www.cnn.com/2004/TECH/biztech/11/10/supercomputerrace.ap/

Again, the estimation of 100 TeraFLOPS doesn't factor in recent brain research (sorry, read it in a science journal months ago and don't have a link) where it was found each neuron is capable of multiple calculations per burst; similiar to a scalar unit (I'm not a hardware engineer, so I might be wrong on the type of hardware I'm thinking of. Could someone correct me if I'm mistaken?). This means that, per burst the neuron can dynamically allocate resources dependent on the type of calculations needed to be done. As a simple example, it might be able to do two additions or subtractions per burst but only one multiplication (as of this year I do not believe scientists have discovered the neuron's exact capabilities).

This estimation also does not factor in the brain's special capabilites for quantum mechanics calculations (again, sorry; can't provide a link). This capability could be compared to a modern CPU's multimedia extensions like MMX, SSE, etc. These extensions allow a CPU to calculate certain mathematical functions much faster than normal. The brain is the same, except it's doing quantum mechanics physics calculations.

The difference between a CPU and the human brain could be compared to the difference between CPUs and graphics processors (GPU). The GPU is built for a dedicated task and while modern GPUs are now capable of calculating general functions like a CPU often times it is far slower; of course it's the reverse for other mathematical functions the GPU was designed for. For those of you still in college or high school, that is why you can't crunch through your calculus homework like crazy...

Finally, this estimation doesn't factor in the possibility of compression, encryption, the fact that the brain is designed to be massively parallel in its computations, and a myriad of other possibilities (quantum computer?). So all in all, I think it's reasonably possible the human brain is capable in excess of 1000 TeraFLOPS.

EDIT: Just realized that CNN link no longer works.

 

Maybe the term "FLOP" isn't exactly right, as those are floating point calculations and i'm not sure the brain "calculating power" can be quantified with floating point maths.

But yeah, we have a hell of a good CPU.

 

I believe emulating the brain with computer hardware is the wrong way of going about things. Just for starters, a computer will probably never be able to compete with brains on a neuron-by-neuron basis; there's just too many of the little buggers and with too many cross-connections.

The brain evolved out of a specific need using tools available to nature. No point in trying to copy it using dead matter... It'll probably just be slow and clumsy and extremely expensive.

 

The brain is a product of evolution. Most things biological resulting from evolution are NOT optimized, but are usually more generic in design and function. I'm sure we will be able to build computers 1000's of times faster then any brain given time.

 

We do already. Try calculating a Doom3 scene 60 times a second with your brain

 

I got a problem with your comment in that although we might not be able to calculate D3 at 60fps, we can still do alot more than any computer that exceeds 1000 or even a 1,000,000 times the computational power.

Also at 60Hz, my eyes hurt looking at the monitor, which kinda does mean we can process 60fps. 75-85 is good though.

US

 

Sony QRIO, dancing and playing golf

 

Then you'd better hurry up and inform biomimetics researchers of this fact for based upon the journals I've been reading they seem to be under the impression that many biological functions are optimized to the extent physics allows (even beyond our technological prowess I might add).

Anyways, bring on the "wetware"...

 

Yes, but they're biological. Because of that, they suffer many limitations.

Again, I'll say, lifeforms are not optimized. They have to be highly adaptable for survival and evolution. If you had to travel 100 miles on a road, would you rather have do it on a horse mustang, or a ford mustang?
I'm sure the horse is an infinitely more complex machine, but I'm also sure the car will get from one point to another much faster. Is a telescope more complex then a human eye? No way! But you don't see far away galaxies using your naked eye.

If will ever build a virtual brain, it will be optimized for it's functions, using things that would NOT EVER have been available in normal biological evolution. May it be silicon, photonics, quantum computing, or whatever....I'm sure we will get there soon enough.

 

Ah, sorry, misunderstood you. Basically you're saying that biological functions are not optimized for the certain applications WE desire. In any case, a while back researchers managed to build a small neuron-based computer that could calculate generalized mathematical functions that were inputted. Through further genetic engineering it might be possible to "grow" our CPUs of the future. Now whether or not it will be worth it from a performance viewpoint by time this is feasible is another matter.

 

Semi-related article to the thread's topic:

http://www.eurekalert.org/pub_releases/2004-12/uom-dmh121704.php

DNA has an information storage density of 1 bit per cubit nanometer; never mind the built-in error minimization functions.

Physicists involved in biology:

http://www.innovations-report.com/html/reports/life_sciences/report-37795.html

 

now this is a very good quote, and this is inevitable.

Like it was in the past, so it will be in the future - a big change, and an unexpected one, whether it will be for further good, eventually it won't... i guess we better enjoy the good times while they last.