Brain, thinking, consiousness - is it similiar to the CPU AI?

[human1111]

Hi all.

Question: Is the human brain's (intelligence, thinking) function like a computer's cpu/AI? In other words is our brain a biological version of the CPU (and cpu an artificial copy of our brain)?

2nd part of the question: Is it possible to build AI that thinks like the human brain? If no, why not?

thank you all!

[brucegdc]

Part 1: The AI researchers WISH that were the case.... CPUs don't program themselves. There are software simulations of neural networks, but none of them come close to the density of connections of the human brain.

Part 2: Possible, yes - in mechanical terms. Feasible, I don't think so. The density of neurons in the brain is pretty high - 200 billion or so, by some estimates (others say about 11 billion). That's a lot of electronic parts.... one neural network circuit (one neuron) has two capacitors, two resistors, and three transistors (http://computing.breinestorm.net/cir...on+edu+auburn/)

Since current manufacturing technology puts about 20 million transistors per square inch, you're talking a chip size about 30,000 square inches (213 square feet), to equal the number of neurons (roughly - this does not include the capacitors/resistors etc). That's a huge chip, and that doesn't include the casing, power, etc, etc, or the interconnection between the neurons (as I recall in the brain, it's a mesh network, which is a lot of connections).

[farhan]

Well on the basic level , a neuron has 5 input 1 out put gates , transistors have 1 input 1 output , digital logic gates in PC have 1-3 input one output . Logic gates can only work in 0 & 1 , while neurons can give & take multiple variations of signals , like 0 , 0.001 , 0.002 ........, 1 . Neurons R placed in a 3-D space while transistors on a 2-d Chip , that severely restricts the number of connections among each of them . So on the hardware level , there is a biiiiiiiiiiig difference .

On the software side , scientists R trying to make a System on Chip IC , that will have huge number of transisters on a chip , & and the software will use the transistors differently for different functions . Its like we use the same hardware ( brain ) to do all the stuff & our software ( Personality/Psyche ) controls the output .

But still machines R used for repeatitive functions . There is a thing in human mindthat is called Creativity/Imagination , that is still difficult to understand & replicate . A brilliant idea comes out of no where . How can U teach a computer to do this ?? Could a computer come up with Realtivity , or Monaliza ??

[human1111]

Originally Posted by farhan

But still machines R used for repeatitive functions . There is a thing in human mindthat is called Creativity/Imagination , that is still difficult to understand & replicate . A brilliant idea comes out of no where . How can U teach a computer to do this ?? Could a computer come up with Realtivity , or Monaliza ??

That is what I am thinking... Increasing raw computational power of a computer may not be enough to make it THINK like we do, it can make it crunch numbers, but not have fuzzy logic, heuristics, problem solving, creative/imagination abilities like we do. So other than us having stronger "cpu" what does our brain has that makes its quality many levels higher than cpu?


Thank you all for replies that you've made.

[dauer]

Why can't we potentially program creativity and imagination into a computer in the distant future? What is creativity and imagination at the neurological level?


Dauer

[Quahom1]

Originally Posted by human1111

Hi all.

Question: Is the human brain's (intelligence, thinking) function like a computer's cpu/AI? In other words is our brain a biological version of the CPU (and cpu an artificial copy of our brain)?

2nd part of the question: Is it possible to build AI that thinks like the human brain? If no, why not?

thank you all!

One major difference between the brain and a CPU is in the way signals are transmitted. Though CPUs can run calculations much faster than the brain, the CPU can only process in serial progressions (one calculation after another, albeit in rapid sequence), wherein the brain can process billions of signals at the same instant.

Just a thought (costing me about 1000 brain cells to think of)

v/r

Q

[Awaiting_the_fifth]

Imagination and creativity are currently far beyond the reach of any computer.

50 years ago, a computer that could create a textured 3D world full of monsters was pure sci fi.

We've only just begun.

And when no one can tell which pieces of art were produced by a human, and which were computer generated, then the really good philosophical debate will begin.

[photon]

The brain stores memory in three dimensions. So far we haven't been able to duplicate that but we will with photonic crystals.

[bob x]

The human brain is designed to magnify small discrepancies from the microscopic level of quantum indeterminacy to the macroscopic level. Neurons contain long-chain molecules sheathed in "microtubules" which shield them from the effects of ambient electric fields, so that a single electron's "random" (that is, "free", not correlated to the distribution of the material particles in the universe) "decision" to jump orbitals or not jump can suffice to lead to the neuron triggering or not triggering and thus to a cascade of effects, ultimately perhaps to you raising your arm or not raising your arm. It would be impossible to trace a naturalist, materialist chain of causation explaining WHY you raised your arm further back than: this electron jumped, "just because".

The computer, on the contrary, is designed to suppress all small discrepancies. The electrical states are divided into classes "1" and "0" with a sufficiently wide enough between them that a single electron's decision to go against the grain cannot make any difference to the subsequent cascade of effects. The upshot is that the computer does what it is programmed to do: you can write a program to simulate a sense of purpose, but if there is a bug in that program, the computer will not think "Aha! this is supposed to be a purposeful program!" but rather will do whatever the bug tells it to do.

We can only build a machine that is conscious when we make the device so "twitchy" at the quantum level that it cannot be programmed (that is, there is no determinism to its behavior): then it won't do what we tell it to do; it will have a "mind of its own".

[I, Brian]

If you're looking for Artificial Intelligence, the one place where development is being pushed very hard commercially is in search engines - Google, Yahoo!, Microsoft, all want to produce a search engine that "understands" the searcher intention, and they see teaching computer networks to understand language as the key to that.

In fact, Microsoft already uses a "neural network" that tries to learn to produce rankings for MSN.

So any IE on current models is likely to be less like a direct attempt to replicate a human brain, as much as a very large processing network in it's own right.

2c.

[Silverbackman]

In relation to religion, isn't it a little contradicting to say that robots can program themselves? Doesn't God or Nature program us? The programing language can be just as complicated as DNA.

At the moment we can create and simulate neorons of living things but it costs too much money to create something even equal to one braincell. Mother Nature somehow does it with far less trouble.

As Awaiting_the_fifth said, We've only just begun! We have a long way to go but we will definatley have the power in the future.

The big question is will these new superintelligent human-like robots have souls? Most likelly they won't, but then it will raise many questions such as do these robots have awareness, and many people believe having an aware consios mind is the same as having a soul.

[bob x]

I am not sure what you mean by "human-like robots". If they are still programmed, they are not "human-like"; if they are not, they will not be "robots". What the computer programmers are creating is interesting, but no "HAL" or "Data" will, or could, arise from current lines of research.