Followers 0

# Theory - ultimate AI, at atomic level

## 61 posts in this topic

[quote name='szecs' timestamp='1313095245' post='4847910']
Anyhoo, I was thinking about this a lot lately. More precisely, I was reading an "article" about alien lifeforms, and lifeforms that are not based on electromagnetism as ours (chemistry), but on gravity, or strong force or weak force. Could a whole galaxy become "intelligent"? What does it mean anyway? The best definition I can think of is the ability to manipulate the environment in "unusual" ways (what the fuck "unusual" means?). At least we could detect that. But how can a galaxy manipulate the environment in an unusual way, yet, obeying the laws of physics?

Sometimes I feel that being "intelligent" and this whole "soul" stuff is about defying these laws in a way. Free will seems to be defiyng the laws of physics. But that can simply be an illusion, because according to the principle I mentioned earlier, the brain cannot model itself accurately.
random rambling is over.[/quote]

You might enjoy [url="http://www.amazon.co.uk/Star-Maker-S-F-Masterworks-Stapledon/dp/1857988078/ref=cm_cr_pr_product_top"]Star Maker by Olaf Stapledon[/url]. it speculates on the various forms of life which could evolve in the universe and how their societies might work - including planets where a symbiotic pair of species is dominant or swarm-based intelligence rules, as well as life composed of galactic bodies. These galactic-scale intelligences are almost morally offended at smaller lifeforms trying so persistently to divert objects from their natural course of motion by applying brute force... The book also imagines life in universes with different physical laws.

On the topic of books, the OP might be interested in [url="http://www.amazon.co.uk/Cyberiad-Stanislaw-Lem/dp/0156027593/ref=sr_1_1?s=books&ie=UTF8&qid=1313106980&sr=1-1"]The Cyberiad by Stanislaw Lem[/url] - it includes a couple of interesting stories about simulating reality, in one case to replicate the Earth in order to fast-forward to the present day and make advances based on the accumulated "knowledge" contained in the simulation.

[size="1"](...Not referral links, I just happen to have read these recently =P)[/size]
0

##### Share on other sites
[quote name='WavyVirus' timestamp='1313107362' post='4848015']
[quote name='szecs' timestamp='1313095245' post='4847910']
Anyhoo, I was thinking about this a lot lately. More precisely, I was reading an "article" about alien lifeforms, and lifeforms that are not based on electromagnetism as ours (chemistry), but on gravity, or strong force or weak force. Could a whole galaxy become "intelligent"? What does it mean anyway? The best definition I can think of is the ability to manipulate the environment in "unusual" ways (what the fuck "unusual" means?). At least we could detect that. But how can a galaxy manipulate the environment in an unusual way, yet, obeying the laws of physics?

Sometimes I feel that being "intelligent" and this whole "soul" stuff is about defying these laws in a way. Free will seems to be defiyng the laws of physics. But that can simply be an illusion, because according to the principle I mentioned earlier, the brain cannot model itself accurately.
random rambling is over.[/quote]

You might enjoy [url="http://www.amazon.co.uk/Star-Maker-S-F-Masterworks-Stapledon/dp/1857988078/ref=cm_cr_pr_product_top"]Star Maker by Olaf Stapledon[/url]. it speculates on the various forms of life which could evolve in the universe and how their societies might work - including planets where a symbiotic pair of species is dominant or swarm-based intelligence rules, as well as life composed of galactic bodies. These galactic-scale intelligences are almost morally offended at smaller lifeforms trying so persistently to divert objects from their natural course of motion by applying brute force... The book also imagines life in universes with different physical laws.
[/quote]

Hmm, thanks for that!
0

##### Share on other sites
As far as I know, neural nets aren't used in game AI : the learning process can be very very long, i think the learning process can't be integrated in games for that reason
[/quote]

They have been used in games before, and learning was done during the testing phase, and the results were just used in the released game. Though I believe that current developers stepped away from the idea, as adjusting to the player has become more important to developers, something which other techniques lend themselves much better for.
0

##### Share on other sites
[quote name='sjaakiejj' timestamp='1313155366' post='4848206']
As far as I know, neural nets aren't used in game AI : the learning process can be very very long, i think the learning process can't be integrated in games for that reason
[/quote]

They have been used in games before, and learning was done during the testing phase, and the results were just used in the released game. Though I believe that current developers stepped away from the idea, as adjusting to the player has become more important to developers, something which other techniques lend themselves much better for.
[/quote]

Yeah, I recall a kind of AI bot for Counter Strike, that had to learn the levels (each of them one by one). One learning was about 20 minutes.Maybe it was similar
0

##### Share on other sites
[quote name='szecs' timestamp='1313160739' post='4848241']
[quote name='sjaakiejj' timestamp='1313155366' post='4848206']
As far as I know, neural nets aren't used in game AI : the learning process can be very very long, i think the learning process can't be integrated in games for that reason
[/quote]

They have been used in games before, and learning was done during the testing phase, and the results were just used in the released game. Though I believe that current developers stepped away from the idea, as adjusting to the player has become more important to developers, something which other techniques lend themselves much better for.
[/quote]

Yeah, I recall a kind of AI bot for Counter Strike, that had to learn the levels (each of them one by one). One learning was about 20 minutes.Maybe it was similar
[/quote]
I believe that may have been John Laird (et al) and his [url="http://sitemaker.umich.edu/soar/home"]S.O.A.R.[/url] technology. Mixed reviews from what little I have heard.
0

##### Share on other sites
A trained neural net is a weighted equation. The two are identical. The difference is in how the weighted equation is developed.
0

##### Share on other sites
[quote name='Dave Weinstein' timestamp='1313167942' post='4848303']
A trained neural net is a weighted equation. The two are identical. The difference is in how the weighted equation is developed.
[/quote]
I've been trying to get that into people's heads for years but the schools keep cranking out people who think just [i]saying [/i]"neural network" is sexy. "Weighted sums" doesn't impress people enough.
0

##### Share on other sites
Hello

[quote name='Dave Weinstein' timestamp='1313167942' post='4848303']
A trained neural net is a weighted equation. The two are identical. The difference is in how the weighted equation is developed.
[/quote]
I've been trying to get that into people's heads for years but the schools keep cranking out people who think just [i]saying [/i]"neural network" is sexy. "Weighted sums" doesn't impress people enough.
[/quote]

For a single layer net such as adaline or perceptron, with the activation function=identity , so we can say that the output is just a 'weighted sum' of the inputs
What about models that don't use the Mc Culloch & Pitts model, like spiking neural networks ? (LaPicque model for instance)

I think that 'weighted equation' is more accurate, at least a better shorcut
0

##### Share on other sites
Hello

[quote name='Dave Weinstein' timestamp='1313167942' post='4848303']
A trained neural net is a weighted equation. The two are identical. The difference is in how the weighted equation is developed.
[/quote]
I've been trying to get that into people's heads for years but the schools keep cranking out people who think just [i]saying [/i]"neural network" is sexy. "Weighted sums" doesn't impress people enough.
[/quote]

For a single layer net such as adaline or perceptron, with the activation function=identity , so we can say that the output is just a 'weighted sum' of the inputs
What about models that don't use the Mc Culloch & Pitts model, like spiking neural networks ? (LaPicque model for instance)

I think that 'weighted equation' is more accurate, at least a better shorcut
[/quote]

I refer to them as "functions with parameters", or "functions with too many parameters for their own good".
1

##### Share on other sites
[quote]
"functions with too many parameters for their own good"
[/quote]

What do you mean ? I don't see the point
0

##### Share on other sites
[quote]
"functions with too many parameters for their own good"
[/quote]

What do you mean ? I don't see the point
[/quote]

An ANN consisting of a single neuron with linear activation function is multiple linear regression, and it is well understood since the times of Gauss. An ANN consisting of a single neuron with sigmoid activation function is logistic regression, and it's also well understood. A multi-layer perceptron has so many parameters that training becomes really hard (e.g., it's hard to avoid getting stuck in local minima), and trying to understand what each parameter does becomes hopeless. That's what I mean by "too many parameters for their own good".
1

##### Share on other sites
[quote name='alvaro' timestamp='1313180231' post='4848400']
An ANN consisting of a single neuron with linear activation function is multiple linear regression, and it is well understood since the times of Gauss. An ANN consisting of a single neuron with sigmoid activation function is logistic regression, and it's also well understood. A multi-layer perceptron has so many parameters that training becomes really hard (e.g., it's hard to [u]avoid getting stuck in local minima[/u]), and trying to [u]understand what each parameter does becomes hopeless[/u]. That's what I mean by "too many parameters for their own good".
[/quote]

Ah ok I understand what you meant, thanks

EDIT :
"avoid getting stuck in local minima" -> that's why a momentum is often added in the delta rule (but it doesn't eliminate the risk, and it's done by adding other parameters .... )

"understand what each parameter does becomes hopeless" -> we generally don't care about what the weight values are (and as you say it's hopeless anyway in a 'big' network)
0

##### Share on other sites
This "simulate everything at atomic level" madness that started with the unlimited detail thing must stop [b]NOW[/b]!

For those saying we have enough processing power: Look at [url="http://en.wikipedia.org/wiki/Folding@home"]Folding@home[/url]; it's attempting to do what you're saying (model at atomic level, although not for AI purposes, but to understand protein synthesis), and it takes thousands of computers running through hours just to simulate ONE NANOSECOND. And last time I check, inteligence improves through time. It takes around two decades for a normal human being in the real world to become intelligent enough.

For those saying we only need to compute those atoms that are revelevant NO! We don't know what's relevant or not; everything is connected with everything. One small tiny miniscule thing leads to another and to another which in results leads to a million chain reactions ultimately causing a huge difference. It's called the [url="http://en.wikipedia.org/wiki/Butterfly_effect"]butterfly effect[/url].
1

##### Share on other sites
I can think of two problems. One, even if we had sufficient processing power, there's the fact that the simulation would have to split time up in frames. That adds a certain amount of error, that would make the rules of that world fundamentally different from those of our world. Two, it presumes that we actually know more about the universe than we really do. We don't have perfect knowledge of physics yet. So we'd make assumptions, and the rules of the simulated world would drift even more.

And a third one that just occurred to me: Say you have preposterous processing power and perfect knowledge of the laws of physics. So you build your world (by the way, are you simulating just the world, or the sun as well? That's a whole other bunch of atoms to think about, and it's not like the sun doesn't have some small impact on our world, so four problems really), and you get your intelligence up and running. And now THEY want to make a computer to simulate the universe. So your computer has to simulate a world with a computer trying to simulate the world. And of course they are successful, so the intelligence that emerges in the simulated computer tries to simulate another world.

So now your computer is simulating a world that contains a computer that is simulating the simulated world, which will contain a computer that can simulate the simulated simulated world. And so on the recursion goes. And you want this to run in real time (in fact, millions of times faster than real time). You're essentially asking the computer to simulate the processing power of infinite computers just like itself. This is a logical impossibility.
1

##### Share on other sites
Incidentally, for those who cite "perfect knowledge of the laws of physics" need to google the "3-body problem" and the "butterfly effect".

On-topic, I should have locked this thread when I had the chance. [img]http://public.gamedev.net/public/style_emoticons/default/dry.gif[/img]
-1

##### Share on other sites
In my very personal opinion trying to create the "Ultimate AI" is a waste of time. I don't mean that AI is a subject that should not be developed
since it's at a great level already( like TOPIO). In order to make it more clear think of a simple thing. What humans have that computers don't? Maybe a lot of things but i'll stick to the subject.
So a human has something called "thought" which it's quite easy to understand, but it cannot be abstracted with any kind of mathematical equation. You simply say "i want to go there" and there may be some
obstacles in the way but you "know" how to bypass them without saying "i have to do this maneuver" or jump, it's done right away without any comparing or calculation. So AI i simply a way to simulate thought with mathematics and equations
but real thought cannot be simulated that way. If you want a real example what i mean just test yourself by putting a simple problem to go somewhere but with obstacles on the way, what you do in order to go there?
And see if you can put down that thought with mathematics.
Although AI may be reach a level quite high enough, the "thinking with calculations" is not the real way of thinking so it's quite limited compared to human thought.
Again this is my personal opinion and i'm open to ideas.

- A programmer
0

##### Share on other sites
[quote name='JohnGreek' timestamp='1314625590' post='4855050']
In my very personal opinion trying to create the "Ultimate AI" is a waste of time. I don't mean that AI is a subject that should not be developed
since it's at a great level already( like TOPIO). In order to make it more clear think of a simple thing. What humans have that computers don't? Maybe a lot of things but i'll stick to the subject.
So a human has something called "thought" which it's quite easy to understand, but it cannot be abstracted with any kind of mathematical equation. You simply say "i want to go there" and there may be some
obstacles in the way but you "know" how to bypass them without saying "i have to do this maneuver" or jump, it's done right away without any comparing or calculation. So AI i simply a way to simulate thought with mathematics and equations
but real thought cannot be simulated that way. If you want a real example what i mean just test yourself by putting a simple problem to go somewhere but with obstacles on the way, what you do in order to go there?
And see if you can put down that thought with mathematics.
Although AI may be reach a level quite high enough, the "thinking with calculations" is not the real way of thinking so it's quite limited compared to human thought.
Again this is my personal opinion and i'm open to ideas.

- A programmer
[/quote]
Just because you/we can't think of a way to "put down that thought with mathematics" doesn't mean it can't be done.
Human mind exists in the physical world and one day we will understand particularly everything about the physical world (at least everything useful) and it will be "put down with mathematics". (it doesn't mean we can model and predict the world perfectly, that's a different and impossible thing).
0

##### Share on other sites
[quote name='JohnGreek' timestamp='1314625590' post='4855050']So a human has something called "thought" which it's quite easy to understand, but it cannot be abstracted with any kind of mathematical equation.[/quote]
Very over-generalized. I wrote a whole bloody book on abstracting thought to mathematical equations. One example:

Two identical objects that you desire. The price of one is $2, the other is$2000. Which will you purchase? I can generally abstract that out with the complex mathematical construct known as the "less than" operator.

[quote]You simply say "i want to go there" and there may be some obstacles in the way but you "know" how to bypass them without saying "i have to do this maneuver" or jump, it's done right away without any comparing or calculation.[/quote]
You are wrong. Just because you aren't [i]aware[/i] of the comparing and calculation doesn't mean that it isn't happening. In fact, there are numerous scientifically tested examples that discern how we [i]do [/i]measure and calculate -- even on a subconscious level. Of course, sometimes our perception or belief systems are incorrect and we may, therefore, choose incorrectly. Still, there are calculations being done.

[quote]So AI i simply a way to simulate thought with mathematics and equations but real thought cannot be simulated that way. If you want a real example what i mean just test yourself by putting a simple problem to go somewhere but with obstacles on the way, what you do in order to go there? And see if you can put down that thought with mathematics.[/quote]
This is what I do, sir. This is what I do. And AI programmers need to be able to do it as well.

I leave you with this comic:
[attachment=5161:equations.png]
-1

##### Share on other sites
I would suggest that an atomic representation of a brain is entirely unnecessary (for the purposes of artificial intelligence) when it is theoretically possible to create a completely functional model of the brain at a cognitive/architectural level. What we lack is a sufficient understanding of said architecture (and, more trivially to a theoretical discussion, sufficient storage). In fact neural modelling seems to be more useful as an aid in neuroscience research than in artificial intelligence research.
0

##### Share on other sites
Just ignoring the obvious scaling problem, another problem with trying to "simulate reality" on an atomic level is that we have no idea how they _actually_work_.
Even less on how we can generalize this in any useful way to simulate them "in bunch"

We have lots of models describing how they seem to move and behave in 4d space, but we know thats just part of what an atom is, the model is incomplete, and no-one knows how it "really works", that's the reason they build things like the LHC.

So even if we did manage to simulate every atom in a brain to our current knowledge, we have no idea if this would produce anything useful or would just fall apart.

And again, just simulating what we _do_ know about atoms/molecules for just a few (thousand) atoms in a protein (as in Folding@HOME) is extremely costly.
Atoms are _not_ governed by "a few simple laws" like balls bouncing around.
0

##### Share on other sites
Incidentally, for those who cite "perfect knowledge of the laws of physics" need to google the "3-body problem" and the "butterfly effect".

On-topic, I should have locked this thread when I had the chance. [img]http://public.gamedev.net/public/style_emoticons/default/dry.gif[/img]
[/quote]

I do not believe you understand either of these problems then.

It's easy to simulate 3 bodies. With a powerful enough computer you could even simulate them accurately. The problem asks for a closed expression for their position(time). This is a mathematical failing. Not a physics failing.

The butterfly effect is exactly the same deal.

The original post was an obvious troll. Not made obvious only by the implication that we could simulate a planet, but that the best way to simulate intelligence would be on the atomic level, rather than the cellular level.

I'd also like to point out to another poster that thinks that machines won't be able to "think" that it is not the machine's failure to think but YOUR failure to understand what thought is.

You think your thoughts are under your control but they are just the deterministic result of the processing of your neural network. Simulating an equivalent neural network in software would have the result of a conscious entity of equal intelligence and sentience as yourself.

(simulating a better one would result in a conscious entity capable of understanding that thought is a deterministic result. not an underlying, driving force)
1

##### Share on other sites
[quote name='sooner123' timestamp='1315836466' post='4860677']
Incidentally, for those who cite "perfect knowledge of the laws of physics" need to google the "3-body problem" and the "butterfly effect".

On-topic, I should have locked this thread when I had the chance. [img]http://public.gamedev.net/public/style_emoticons/default/dry.gif[/img]
[/quote]

I do not believe you understand either of these problems then.

It's easy to simulate 3 bodies. With a powerful enough computer you could even simulate them accurately. The problem asks for a closed expression for their position(time). This is a mathematical failing. Not a physics failing.

The butterfly effect is exactly the same deal.

The original post was an obvious troll. Not made obvious only by the implication that we could simulate a planet, but that the best way to simulate intelligence would be on the atomic level, rather than the cellular level.

I'd also like to point out to another poster that thinks that machines won't be able to "think" that it is not the machine's failure to think but YOUR failure to understand what thought is.

You think your thoughts are under your control but they are just the deterministic result of the processing of your neural network. Simulating an equivalent neural network in software would have the result of a conscious entity of equal intelligence and sentience as yourself.

(simulating a better one would result in a conscious entity capable of understanding that thought is a deterministic result. not an underlying, driving force)
[/quote]

If you close your eyes and think of a person or picture...what is it youre seeing? That "picture" that you're imagining is not a picture...its a thought, but one that we interpret as a picture by using different parts of our brain to form it.

When you create an AI...how would they see that "thought" If you programmed them with knowledge of a particular item, could they use their programmed knowledge to picture the item without seeing it?

Makes me think of language...language is actually a barrier that slows our thought process down. If everyone was equally intelligent, perfect beings, we would have no need for language. It wouldnt be telepathy, it would be knowing the answer because its the right thing to do. If we had to communicate with people, we would instantly understand what they needed without exchanging words because we could interpret the need without having to talk.

its like a team game..either digital or athletic. you become a cohesive unit...multiple brains melding into one to the point where you can predict what the other is doing without talking.

0

##### Share on other sites
[quote name='fr0st2k' timestamp='1315841994' post='4860710']
[quote name='sooner123' timestamp='1315836466' post='4860677']
Incidentally, for those who cite "perfect knowledge of the laws of physics" need to google the "3-body problem" and the "butterfly effect".

On-topic, I should have locked this thread when I had the chance. [img]http://public.gamedev.net/public/style_emoticons/default/dry.gif[/img]
[/quote]

I do not believe you understand either of these problems then.

It's easy to simulate 3 bodies. With a powerful enough computer you could even simulate them accurately. The problem asks for a closed expression for their position(time). This is a mathematical failing. Not a physics failing.

The butterfly effect is exactly the same deal.

The original post was an obvious troll. Not made obvious only by the implication that we could simulate a planet, but that the best way to simulate intelligence would be on the atomic level, rather than the cellular level.

I'd also like to point out to another poster that thinks that machines won't be able to "think" that it is not the machine's failure to think but YOUR failure to understand what thought is.

You think your thoughts are under your control but they are just the deterministic result of the processing of your neural network. Simulating an equivalent neural network in software would have the result of a conscious entity of equal intelligence and sentience as yourself.

(simulating a better one would result in a conscious entity capable of understanding that thought is a deterministic result. not an underlying, driving force)
[/quote]

If you close your eyes and think of a person or picture...what is it youre seeing? That "picture" that you're imagining is not a picture...its a thought, but one that we interpret as a picture by using different parts of our brain to form it.

When you create an AI...how would they see that "thought" If you programmed them with knowledge of a particular item, could they use their programmed knowledge to picture the item without seeing it?

Makes me think of language...language is actually a barrier that slows our thought process down. If everyone was equally intelligent, perfect beings, we would have no need for language. It wouldnt be telepathy, it would be knowing the answer because its the right thing to do. If we had to communicate with people, we would instantly understand what they needed without exchanging words because we could interpret the need without having to talk.

its like a team game..either digital or athletic. you become a cohesive unit...multiple brains melding into one to the point where you can predict what the other is doing without talking.

[/quote]

Best advice, stay away from metaphysics or philosophy when working AI as they are not based in fact (not to be confused with conceptual facts in philosophies) or more importantly are not grounded in reality, which is where you have to make the AI funciton. Additionally, best to avoid the "perfect emulation" of thought or thinking and go for "best approximation" as perfect emulation requires magnitudes of processing power above the "best approximation".

Case and Point: SNES emulation can be perfectly emulated under existing hardware, however, it only has recently been possible to do so, mainly due to the amount of processing power required to perfectly emulate SNES hardware.

Another Example: IBM Watson. Watson is as dumb as a box of rocks, but appears intelligent due to the way it verifies and assembles evidence information based on the question asked and then ranks the results. Watson isnt a perfect emulation of question-answering. It is a current "best approximation", and even then Watson still takes 2800+ processors and Terabytes of Secondary Memory (RAM) to answer basic questions in 2-6 seconds (several hours on a lone processor).

Not saying to lower hopes or expectations, but to properly set them within the confines of what is possible. Build and reach from there. Serendipity is rare, Progress is unavoidable.
0

2