Memristor Minds, the Future of Artificial Intelligence

godlessgambler writes "Within the past couple of years, memristors have morphed from obscure jargon into one of the hottest properties in physics. They've not only been made, but their unique capabilities might revolutionize consumer electronics. More than that, though, along with completing the jigsaw of electronics, they might solve the puzzle of how nature makes that most delicate and powerful of computers — the brain."

Comments

Oblig. wiki-link

[Eudial]

What the hell is a memristor, you ask?

Re:Oblig. wiki-link

[Requiem18th]

The first place being xkcd

Re:Oblig. wiki-link

[calzakk]

I found the link useful :-)

Instead of going to Wikipedia and typing the word into the search box and hitting Go... I just clicked the link!

Re:Oblig. wiki-link

[General Wesc]

I pressed ^t, typed in 'wp memristor' (actually it was either 'wp ^v' or 'wp [middle-click]'), and pressed Enter. Bookmark keywords are your friend.

This is Slashdot. I think we should assume people are capable of locating Wikipedia on their own.

Re:Oblig. wiki-link

[tenco]

It's called "adding context" and should have been in the summary, anyway. It's called HTML for a reason.

Re:Oblig. wiki-link

[Anonymous Coward]

Boy, that never happens around here.

Re:Oblig. wiki-link

[linhares]

I'm new here. What's this "Wikipedia" and where do I find it? Here. That'll sort this whole thread out...

He asked for an explanation. He may not yet know how to click on a link. Wikipedia is a free,[5] web-based multilingual encyclopedia project supported by the non-profit Wikimedia Foundation. Its name is a portmanteau of the words wiki (a technology for creating collaborative websites, from the Hawaiian word wiki, meaning "quick") and encyclopedia. Wikipedia's 13 million articles (2.9 million in the English Wikipedia) have been written collaboratively by volunteers around the world, and almost all of its articles can be edited by anyone who can access the Wikipedia website.[6] Launched in January 2001 by Jimmy Wales and Larry Sanger,[7] it is currently the largest and most popular general reference PENIS PENIS PENIS PENIS

Re:Oblig. wiki-link

[Eternauta3k]

This is Slashdot. I think we should assume people are capable of locating Wikipedia on their own

This is slashdot. I think we should assume people are too lazy to search wikipedia (even with a shortcut), let alone RTFA.

I'm always taken back by this

[msgmonkey]

That we've developed a whole industry based on an incomplete model, I wonder how things would have developed if the memristor had existed 30 years ago. Exciting times as a lot of things will be re-examined.

Re:I'm always taken back by this

[Anonymous Coward]

Probably nothing significant, seeing as you can emulate exactly what a digital memristor does with 6 transistors and some electricity always applied. Memristors in CPU/logic would not be viable because of their low wear cycles and very high latencies. It would make for some nice multi-terabyte sized USB sticks though.

As for its analog uses, Skynet comes to mind...

Re:I'm always taken back by this

[jerep]

we've developed a whole industry based on an incomplete model

Wait you mean this is the first time this happens? I thought schools were the first to do that.

Re:I'm always taken back by this

[Marble1972]

Probably nothing significant, seeing as you can emulate exactly what a digital memristor does with 6 transistors

Exactly right.

It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required. Faster computers might be nice - but it'll always comes down to the algorithm.

And actually the sooner we create Skynet - the better the chance we have to beat it. Because if we wait too long - that super fast hardware it will be running will could make it too hard to beat. ;)

Re:I'm always taken back by this

[peragrin]

Of course if you currently multiply the 100 million or more transistors in a current cpu by 6 you don't have any kind of problem do you? Of course a memresistor is closer in design to a permanent RAM Disk. You can turn off the system as much as you want but it instantly restores you right from where you left it.

Now that it is proven all that matters is figuring out how best to use it and what limitations it has.

Re:I'm always taken back by this

[Yvanhoe]

No. This is a lot of gross overexageration.
Our computers are Turing-complete. Point me to something that is missing in this before I get excited. This new component may have great applications, but it will "only" replace some existing components and functions. It is great to have it but it is nothing essentially missing.

Re:I'm always taken back by this

[madkow]

And actually the sooner we create Skynet - the better the chance we have to beat it. Because if we wait too long - that super fast hardware it will be running will could make it too hard to beat. ;)

Or the better chance we have to learn live with it. James Hogan's 1979 book "The Two Faces of Tomorrow" details a plan to deliberately goad a small version of a self aware computer (named Spartacus) into self defense before they built the big version. When Spartacus learned that humans were even more frail than he and equally motivated by self preservation he chose to unilaterally lay down arms.

Re:I'm always taken back by this

[Requiem18th]

I don't know, with a 10,000 write limit If my brain was made of memristors I'd be terribly mortified.

Re:I'm always taken back by this

[Hurricane78]

Except if what I and many other people think is true: That the only difference between our spiking neural nets and Skynet is the processing power.

Re:I'm always taken back by this

[LuxMaker]

And actually the sooner we create Skynet - the better the chance we have to beat it. Because if we wait too long - that super fast hardware it will be running will could make it too hard to beat. ;)

And why did this suddenly make the tune "Just beat it" play in my mind? Michael Jackson vs. Skynet?

Re:I'm always taken back by this

[4D6963]

the only difference between our spiking neural nets and Skynet is the processing power.

No, repeat after me: Putting a shitload of neural networks on a supercomputer won't create a strong AI.

A very common persistent misconception.

Re:I'm always taken back by this

[TapeCutter]

"It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required."

On the contray, I think you need a algorithmic breakthrough to understand the brain but you don't need a new algorithim to create a brain. Humans have built and used many things well before they had a theoretical basis for how they worked, for example people were using levers to build pyramids long before archimedes came and gave us the "lever algorithim".

Re:I'm always taken back by this

[ScrewMaster]

Skynet spreads onto almost every computer in the world (ie. there is no central core).

That was just the rubbish from the third movie, which film I'm personally trying to forget. It was also ridiculous: once Skynet nuked everyone and in the process shut down all power distribution and communications networks, all those millions of computers running some little bit of Skynet would be turned off and isolated anyway.Killings us would have been instant suicide for Skynet.

The original 80's vision of Skynet as a vast artificial intelligence living in a cavern somewhere running on its own power supply still makes a lot more sense.

Re:I'm always taken back by this

[GigaplexNZ]

Memristors in CPU/logic would not be viable because of their low wear cycles and very high latencies.

That's a current manufacturing limitation, not something inherent to what a memristor is. Had these been discovered much sooner, we would be much better at manufacturing them and they probably would have made a significant impact.

Re:I'm always taken back by this

[Sponge Bath]

This talks about neuronal replacement. It looks like your brain may have a write limit, it just automatically replaces worn out bits.

Re:I'm always taken back by this

[Burnhard]

it's an algorithm breakthrough that's required

That is, of course, making the assumption that intelligence doesn't require Consciousness and that Consciousness can be captured in an algorithm. Two somewhat dubious pre-requisites.

Re:I'm always taken back by this

[divisionbyzero]

Right. AI is a software problem, not a hardware problem. That's not to say that current hardware could run the software should it ever be devised, but once we know what the software is we can build the hardware that will run it. So, how do we come up with the software if we don't have the hardware to run it? It's called philosophy.

Re:I'm always taken back by this

[The End Of Days]

Why dubious? There's no evidence either way, so poo-pooing the possibility doesn't make you smarter, it just makes you close-minded.

Re:I'm always taken back by this

[ceoyoyo]

You're still thinking about simulating intelligence using a standard computer, in which case you're right, you need the right algorithm.

What they're proposing is not to simulate a brain but to build one. There is no algorithm. It might be sensitive to how you wire things up, but probably not excessively so, otherwise it would be very difficult to evolve working brains. The key is getting the right components to build the thing out of.

Re:I'm always taken back by this

[ceoyoyo]

"Repeat after me" is really annoying. If you're going to be that irritating you'd better have some pretty strong evidence to back yourself up. Where is it?

Re:I'm always taken back by this

[Twinbee]

How about infinite CPU speed (+ memory). Would that help?

Re:I'm always taken back by this

[Burnhard]

so poo-pooing the possibility doesn't make you smarter, it just makes you close-minded.

As there's no evidence either way, metaphysical doubt, it seems to me, is quite a sensible position.

Re:I'm always taken back by this

[4D6963]

You want evidence that something infeasible (assuming you won't consider that we currently can't put enough neural nets in a supercomputer powerful enough to satisfy your demand) is impossible? I want some evidence that filling a 1 billion cubic meters bag with spiders won't turn into an evil blob-like insect overlord. Where is it?

However if by 'evidence' you mean 'reason', then I'll tell you that researchers spent decades putting ever increasing shitloads of neural networks on ever increasingly powerful supercomputers and that nothing particular ever came out of it, and that neural networks are just algorithms anyways, so throwing a whole bunch of power into it won't solve it, as we know what neural networks can or cannot do. If neural networks could possibly create some sort of strong AI, even if it was only a bug-like AI, we'd know it by now. Don't be fooled by the association made with actual neurons. If you're interested in AI because of HAL 9000 and the likes, then don't get into the field of AI, because there's nothing but ever-lasting disappointment for you.

Strong AI is to computer research as instant matter teleportation or (backwards) time travel are to astrophysics, nice in scifi, but never gonna happen.

Re:I'm always taken back by this

[Not Summer Glau]

I still have John Connor on speed-dial.

Oh yeah? Well, I have Cameron's number!

Re:I'm always taken back by this

[ceoyoyo]

Ah yes, the "our computers are incredibly powerful and we've tried it and it didn't work so the whole class of solutions is obviously ruled out" argument.

Before you make (extremely condescending) statements that something is impossible, you should at least make sure you qualify your terms properly.

"I think it's very unlikely that using current neural network algorithms on computers with current or near future capacities will produce a strong AI" would be a good start.

We certainly do not know what the limits of "neural networks" (as a general class of algorithms) are. We also don't have anything like the computing power to properly simulate a neural network with a capacity where we'd expect to see "intelligence."

You might be correct. Then again, you may well not be. Even if you are, the only people who will listen to posts like yours are people who already agree with you.

Re:I'm always taken back by this

[Jack9]

How do you know we don't have the computing power? You're assuming that that's the barrier. "We'd expect to see intelligence" sounds an awful like you have a pre-conception of what you want to see. I'd love to subscribe to your newsletter.

Re:I'm always taken back by this

[iluvcapra]

I have Kathyrn Bigelow's. We should get them back together....

Re:I'm always taken back by this

[ceoyoyo]

If you're trying to simulate a brain you have a pretty good idea of how many neurons and synapses are in your simulation, and how closely their behaviour matches those of the real thing. Generally there's a tradeoff - you can make your simulation more realistic at the expense of simulation fewer elements. There was a story a while ago about an effort to very accurately simulate a neuron on a supercomputer. One neuron.

Now, for whatever definition of "intelligence" you have, you can probably find a brain in nature that exhibits approximately that level of behaviour. For some people the bar is very low, say, cockroach level. That level of "intelligence" has already been demonstrated in simulations.

Now that you know the natural equivalent, you can measure the size and sophistication of your simulated brain and compare it to the size of your natural analogue. If your simulated brain is much smaller than your natural analogue, you probably shouldn't expect to see behaviour indicating your chosen definition of intelligence. If it is larger, and you still don't see it, you should examine the accuracy - perhaps you've over simplified something critical.

If you set the bar low for intelligence and are happy with cockroach behaviour, I have great news - we've already reasonably achieved it. Since the poster I was replying to is adamant that we have NOT achieved artificial intelligence in a computer simulation, clearly he sets the bar higher. I was using his implied definition when I said "we'd expect to see intelligence."

Re:I'm always taken back by this

[The End Of Days]

I guess I prefer to stay on the positive side since no one ever made progress by deciding things were impossible and giving up. Always surprised to see people not getting that. I realize that's my silliness though.

Re:I'm always taken back by this

[Lord Kano]

It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required. Faster computers might be nice - but it'll always comes down to the algorithm.

Fast enough computers will allow us to develop algorithms genetically. Come up with a set of parameters and let evolution do the job for you.

LK

Re:I'm always taken back by this

[HiThere]

Your argument doesn't sound designed to convince anyone except yourself. If that's your goal, then congratulations, you appear successful. Most people, however, choose a slightly different goal.

To arrive at a simulation of the truth, however, it's generally necessary to define the terms of importance to the discussion and not to overgeneralize. E.g., "neural network" is not a generic term, it's a category, with many differing entities. Many of those differing entities have very different learning patterns. And the magnitude of differing varieties of neural networks is probably as large as (or larger than) all the programs that have ever been written to date. (Rough estimate, and counting all the development patterns from each neural network as being the same network.)

Then there's the question of what's intelligent, and how you would recognize it... Please remember that intelligence generally needs a favorable environment to develop in, or it fails to develop. But what's favorable depends on the details of the learning pattern of the network.

Up to date there have been very few attempts to create a genuinely intelligent entity. The few failures aren't really very significant, except as they are guideposts to where not to make some mistake (usually one obvious in retrospect).

FWIW, I, personally, believe that the amount of intelligence required to interact with people has been grossly overestimated. What I believe to be needed is not primarily intelligence, but rather an extremely large database of "how the world works" and "what's happening here and now". I don't happen to believe that neural nets are the most computationally efficient means of creating intelligence, but they are an effective means of categorizing unstructured data. I think they were chosen because they ere the easiest computational structure to evolve, and once evolved they were "good enough" that no competitor ever became better quickly enough to surpass them.

Re:I'm always taken back by this

[v(*_*)vvvv]

That we've developed a whole industry based on an incomplete model

In hindsight, every industry ever is an incomplete model.
We will always have much to look forward to.

Re:I'm always taken back by this

[cerberusss]

You're saying if it's some kind of magic incantation.

Do you know what genetic algorithms are? It's just a bunch of data, plus a "fitness" function in a loop of say 100.000 runs or whatever, and (more or less) randomly changes the data to see if the fitness function gives a better result. First you'll have to write the fitness function, then you'll have to think about how you randomly change the data, and then you'll have to manually tweak it.

It's a simple algorithm, which is only usable in special cases. You're basically saying: see, if we have better engines, and we have automatic transmissions, then we'll have self-replicating cars.

Re:I'm always taken back by this

[v(*_*)vvvv]

It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required. Faster computers might be nice - but it'll always comes down to the algorithm.

I fear this may be an oversimplification. Current software can simulate virtually any hardware, be it existing or hypothetical. We also have no problem simulating physics and doing math. These memristors may be novel pieces of hardware, but simulating them seems quite trivial, as computers already have a "perfect" memory.

However, the ability to physically implement such a device seems why this is so important, and I cannot say for certain that hardware inventions will not contribute to the evolution of AI.

Algorithms drive hardware. Sure, new algorithms on current hardware will lead to breakthroughs, but we aren't even close enough to know if that is indeed enough to solve the problem. For if we even knew exactly what the problem was, we would already be much closer to solving it.

Re:I'm always taken back by this

[linhares]

IF I HAD MOD points, you'd all be getting offtopic! --skynet

Re:I'm always taken back by this

[Burnhard]

I guess I prefer to stay on the positive side since no one ever made progress by deciding things were impossible and giving up. Always surprised to see people not getting that. I realize that's my silliness though

You don't make progres by randomly flailing around or by applying erroneous principles to the problem at hand.

Re:I'm always taken back by this

[technofix]

All models are incomplete. If they weren't simplifications they'd be useless as models. There are now good alternatives to models but that's another story. Ask if you want to know.

Re:I'm always taken back by this

[adamgolding]

"It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required. Faster computers might be nice - but it'll always comes down to the algorithm. "

If you mean *strong* AI, a better algorithm may, in fact, create strong AI, but we'd never know it, until we can learn under what conditions a system would be conscious.

Re:I'm always taken back by this

[iluvcapra]

You're not skynet. I am.

Re:I'm always taken back by this

[gronofer]

I don't know, with a 10,000 write limit If my brain was made of memristors I'd be terribly mortified.

Try not to be so indecisive.

Re:I'm always taken back by this

[4D6963]

Up to date there have been very few attempts to create a genuinely intelligent entity.

Yeah, why would that be? Perhaps because no one has any fucking clue how to go about doing this?

Re:I'm always taken back by this

[An Onerous Coward]

It's possible to simulate a mouse's brain today. That's about 1 gram of brain matter. The human brain weighs about 1350g. Eleven more doublings, maybe throw in a couple more (under the assumption that the simulations need to be more thorough), and you're pretty well there. So before 2030 would be a good guess.

I suspect that, rather than deriving Artificial Intelligence from first principles, we'll just start by trying to map and run the human brain. He's right: it's not enough to pour neurons into a simulator. You have to wire them up in very specific configurations, which I don't think we'll discover without guidance from existing examples.

Just as an aside, has anyone ever read a New Scientist story where the subject was anything less than An Astonishing Breakthrough That Will Change Everything Forever? I'm becoming ever more convinced that they suck.

Re:I'm always taken back by this

[HiThere]

Largely because everybody knows that the computers are underpowered. Progress is being made on pieces of the job. (You didn't think "intelligence" was unitary did you?)

Logical reasoning was easy. Also trivial. Also an extremely small piece of what was needed. (Bayesian probability estimation is interesting, but it's not deeply intelligent.) There's some fancy pattern matching that needs to be implemented, etc. Exactly which pieces are missing depends on your approach. I don't like neural networks because it's too difficult to understand how they're processing the data, but they're needed in certain areas because they can figure out how to classify data without foreknowledge. But you don't want to use them in large areas because they're too inefficient. (They'd probably up the required computations by an order of magnitude...but that's a really wild guess, and others disagree.)

Several people claim to know how to build a general purpose AI, if only they could afford to. And they aren't all impecunious. Massive computation is expensive. And then there's the matter of how to train the AI once you've got it built. Is a simulated world rich enough? The answer isn't clear, but if you've got to build a physical body, then the costs go up by at least double, and probably lots more. So does the time involved.

Personally, I've given up on creating a real AI. I know I'm years behind literally dozens of other people that I know of, and I also don't have the financial resources necessary. Computers are getting cheaper per calculation, but I could stretch to afford something powerful enough to host a chicken equivalent AI...or possibly a sparrow equivalent. (It wouldn't be doing the same things, and I'm presuming that my approach would be lots more efficient than a neural net. So comparisons are only to give one a very rough idea or what I mean. I'm hoping for reasonable language processing, but not for much understanding...not that I've gotten there yet.)

Re:I'm always taken back by this

[4D6963]

Several people claim to know how to build a general purpose AI, if only they could afford to.

Several you say mmmh. The real question is, do they outnumber the people who claim to know how to build a perpetual movement machine if they had the means?

Re:I'm always taken back by this

[ceoyoyo]

"He's right: it's not enough to pour neurons into a simulator. You have to wire them up in very specific configurations...."

Our only examples imply otherwise. Suppose we want to achieve human-style intelligence. Unless you're a creationist, it's very likely that the human brain (and all other brains) is the product of an evolutionary process. This implies that the human brain can be produced by a series of gradual refinements of simpler brains. Intelligence cannot be an all or nothing phenomenon that is critically sensitive to the precise arrangement and interconnection of billions of component neurons and synapses. The brain cannot be irreducibly complex.

It is possible that intelligence requires some sort of uncomputable phenomenon, in which case conventional computer simulations will always fail, but physical models will succeed.

Re:I'm always taken back by this

[Nikola Tesla and You]

Probably nothing significant, seeing as you can emulate exactly what a digital memristor does with 6 transistors and some electricity always applied.

Except 1 memristor = (40nm)^3, which is considerably smaller than 1 modern solid state transistor, not even considering 6 of them to achieve the same functionality. Thus, higher density + higher functionality + lower power usage = uber pwnage

Re:I'm always taken back by this

[Handlarn]

Why wouldn't consciousness be fit to be made into an algorithm? You think somehow consciousness doesn't follow the same rules of physics as everything else in the universe?

Re:I'm always taken back by this

[Burnhard]

Why wouldn't consciousness be fit to be made into an algorithm? You think somehow consciousness doesn't follow the same rules of physics as everything else in the universe?

Frankly, no, I don't. Consciousness is not supervenient on the physical, so there's no reason to believe that you can code an algorithm for it (how?). It does correlate with physical mechanism (hence neural correlates of Consciousness), but otherwise the connection cannot be studied, i.e. equations cannot be written concerning Consciousness. When physicists have explained all of the fields, forces and particles in the Universe, they will still have something left to explain.

Now that is not to say that some device cannot be made conscious (our brains are, after all). It is, however, to say that the action of such a device must be capable of generating the physical effects that are the correlates of consciousness. An algorithm (a model), does not behave in such a manner in itself. With an appropriate physical mechanism, it might.

Re:I'm always taken back by this

[metaforest]

The human brain weighs about 1350g.

While so far as I have read most of the mass of the brain is not composed of active neurons, though IIRC the supporting structures do directly influence the processing of the active layer.

Additionally you left out about 75g of CNS that does a lot more "in-line" processing than AI geeks seem to acknowledge.

I couldn't find any reference that estimated the active mass of the brain. It might be more telling to compare active masses rather than total brain mass.

To me this is a bit like comparing engine performance by measuring total mass of a complete engine rather than the more meaningful (though still kind of iffy) displacement.

Re:I'm always taken back by this

[An Onerous Coward]

I'm not disagreeing with you. But I do think that the most straightforward approach* to a human-like intelligence is to work from a known good pattern. Reimplementing evolution seems likely to be a long, drawn out process, and one which is unlikely to lead to an intelligence "like us". I think it's important that artificial intelligences be a lot like us.

* Not necessarily the fastest approach, or the most rewarding. Just the most certain to succeed.

Re:I'm always taken back by this

[ceoyoyo]

Sure, I think a good approach is to use the model we have. But I don't think we need to sweat the small stuff. The precise arrangement of connections cannot be all that important.

Two other sources of evidence for this: 1) there's not enough capacity in the sperm and egg to transfer a detailed connection-by-connection brain map, and 2) brain plasticity. You can do a lot of damage to a brain and still have it work. When the brain is young you can even do things like hemispherectomies and still have it function very well.

Re:I'm always taken back by this

[Mr. Mikey]

Compared to electronics, neurons are large and slow. They're made of plain old matter.

So, tell me, what is this insurmountable barrier that prevents us from building a piece of hardware that does what brains do? We don't know how to do it now, you speak as if you have certain knowledge we'll never be able to.

So, what is this certain knowledge?

Re:I'm always taken back by this

[4D6963]

So, tell me, what is this insurmountable barrier that prevents us from building a piece of hardware that does what brains do? We don't know how to do it now

The fact that we don't know how to do it now (and by this I mean, we don't even remotely have any clue). Arguing that something we don't even remotely how it could be done is possible is pretty much arguing that anything is possible.

Electrical Memristors Don't Exist Yet

[indigest]

From the article:

What was happening was this: in its pure state of repeating units of one titanium and two oxygen atoms, titanium dioxide is a semiconductor. Heat the material, though, and some of the oxygen is driven out of the structure, leaving electrically charged bubbles that make the material behave like a metal.

The memristor they've created depends on the movement of oxygen atoms to produce the memristor-like electrical behavior. Purely electrical components such as resistors, capacitors, inductors, and transistors only rely on the movement of electrons and holes to produce their electrical behavior. Why is this important? The chemical memristor is an order of magnitude slower than the theoretical electrical equivalent, which no one has been able to invent yet.

I think the memristor they've created is a great piece of technology and will certainly prove useful. However, it is like calling a rechargeable chemical battery a capacitor. While both are useful things, only one is fast enough for high speed electronics design for applications like the RAM they mentioned. On the other hand, a chemical memristor could be a flash memory killer if they can get the cost down (which I doubt to happen any time soon).

Re:Electrical Memristors Don't Exist Yet

[j0hnyquest]

I completely agree. Right now, the memristor is more of something theoretical physicists splooge over rather than something that will be anywhere near useful in the foreseeable future. I believe its like a lot of the quantum innovations currently being worked on (read IEEE potentials, there are a bunch)... lots of potential, but nothing concrete yet.

Re:Electrical Memristors Don't Exist Yet

[Hurricane78]

How about an optical memristor?

Why focus on hopefully soon outdated technology. :)

Re:Electrical Memristors Don't Exist Yet

[tennin]

The charge carriers are still electrons. Oxygen vacancy migration tunes the distance the electrons must tunnel. At the extremely small scale where memristance becomes dominant, the speed of ion diffusion seems negligible.

Re:Electrical Memristors Don't Exist Yet

[Velocir]

Mod parent up, +Insightful!

Tha's goint to be the NEXT BIG THING

[12357bd]

in the computer world.

The question is: will be see the result in our lives?

I really wish so, but the succes has stalled computer innovation. Thirty years ago we expected to be able to talk to our machines, now those advances can make it finally possible. Will the industry and economics be able to adapt to make it possible in our life time frames?

Re:Tha's goint to be the NEXT BIG THING

[Yvanhoe]

AI needs new algorithms to progress. Electronics will not change the way we program computers. They are already Turing complete, a new component adds nothing to the realm of what a device can compute. Expect a revolution in electronics, but IT people will not see a single difference (except maybe a slight performance improvement)

Re:Tha's goint to be the NEXT BIG THING

[avg_joe_01]

Turing complete

You keep using that word. I do not think it means what you think it means.

Re:Tha's goint to be the NEXT BIG THING

[12357bd]

Old designs were not fully explored, ie: Turing's 'intelligent or trainable' machines. This kind of electronics can do those old concepts viable, that's IMO the NEXT BIG THING, not just algorithms (looped circuitry is not hard to simulate, is hard to predict).

The Von newman architecture of our 'computers' was just one possibility, not the only or the best, just the convenient. New hardware processing habilities, could lead to new kinds of maybe not 'programable' in the current sense of the word, but 'trainable' machinery.

Re:Tha's goint to be the NEXT BIG THING

[4D6963]

but the succes has stalled computer innovation

No, reality got in the way. As much as you can want to have a HAL 9000 in your computer, it's not going to happen, because as far as we know it might just be theoretically impossible to create something like that.

Thirty years ago we expected to be able to talk to our machines, now those advances can make it finally possible.

No it's not. What makes you think it's gonna help with anything you talk about? That's typical of throwing the word "neuron" into a technology story, just as soon you have a bunch of readers peeing their pants fantasising of HAL 9000/Skynet/whatever else you people think is a cool scifi example of strong AI.

Re:Tha's goint to be the NEXT BIG THING

[4D6963]

Huh??! Please explain how this was used incorrectly in the GP post, and then explain what it means to you.

Re:Tha's goint to be the NEXT BIG THING

[Baron_Yam]

I can point to a common example of a machine that produces intelligence... the human brain.

So, given that nature did it once I'm confident it's not theoretically impossible.

Re:Tha's goint to be the NEXT BIG THING

[4D6963]

A machine is a device. A device is a human invention. Men didn't invent brains.

Besides by theoretically impossible I was talking about doing it algorithmically.

Re:Tha's goint to be the NEXT BIG THING

[ceoyoyo]

Our computers are not Turing complete. To be so, they would require infinite memory and infinite time.

"Turing complete" is one of those things know-it-alls say like "correlation does not imply causation." The vast majority of statements on Slashdot that contain those phrases are made by people who do not have more than a very superficial knowledge of either one.

Which is a pity in this information age. Even Wikipedia has both right: Turing Complete, Correlation Does Not Imply Causation.

Re:Tha's goint to be the NEXT BIG THING

[Baron_Yam]

You're talking religion, not science.

The brain is an electro-chemical machine 'designed' by random chance controlled by natural selection.

The algorithms are in there, and there is no reason to believe they can't be copied by man or even that we might figure them out ourselves.

If the algorithms were theoretically impossible then your brain wouldn't exist.

Re:Tha's goint to be the NEXT BIG THING

[4D6963]

I stand corrected and withdraw my GP comment!

Re:Tha's goint to be the NEXT BIG THING

[4D6963]

That supposes that the entire universe is algorithmically reproducible, which is a view that holds some merit, however I still doubt we will manage to come up with anything resembling that fabled 'strong AI'. At least I can live in the comfort of doubtlessly living to not being proven wrong.

Re:Tha's goint to be the NEXT BIG THING

[Yvanhoe]

Please explain to me how this component extends the range of process we can simulate/predict. I fail to see the link with the Turing's publication you mention that proposes concepts that are completely implementable with a current CPU.
This is only a component with a new electrical behavior, for heaven's sake ! Completely simulatable, its behavior is linear, I fail to see how it could have profound implications. It will maybe simplify some electric circuits that needed 3 capacitors and a coil, but it won't change a single thing in the way we program computers.

Re:Tha's goint to be the NEXT BIG THING

[Yvanhoe]

From Turing completeness :

"While truly Turing-complete machines are very likely physically impossible, as they require unlimited storage, Turing completeness is often loosely attributed to physical machines or programming languages that would be universal if they had unlimited storage. All modern computers are Turing-complete in this loose sense, or more precisely linear bounded automaton-complete."

Please point me to a problem unsolvable by a Turing-complete machine (you can choose the definition you prefer) that could be solved by the addition of memristors. Failing that, I would still be interested in a way to make NP problems solvable in polynomial time (the promise of quantum computing). Until then, I will just dismiss it as an over-hyper electronic component with only small niche application and a need for new investors.

Re:Tha's goint to be the NEXT BIG THING

[monoqlith]

No. I don't think the solution will be algorithms running on existing digital electronics.

Our brain is an analog machine. Its plasticitiy is not limited to two discrete states. Therefore, the 'software running on hardware' model for how intelligence works is not the most efficient explanation. Our brains operate the way they do because of they way they are organized, not because they are programmed in the sense we usually understand it. To put it another way, the software 'instructions' (algorithm) and the information processing(the processor) are really the same component in our intelligent machine. The mind is not software running on the brain - it *is* the brain.

In my opinion, the first artificial mind will be structured in a similar way, not in the processor-memory paradigm, but in the 'learning circuit' model. This is so because we have 'prior art' (from nature) that demonstrates that this already works for creating an intelligent machine - us.

Re:Tha's goint to be the NEXT BIG THING

[Will.Woodhull]

AI needs new algorithms to progress.

That's quite an assertion. But it could also be the case that we have all the algorithms we need to produce a sentient machine that would then independently develop its own intelligence. But that we simply have not been putting the different algorithms together in the correct fashion.

Maybe the box of Legos we've got already has all the pieces we need to build this cathedral, and we just need to learn to put them together to make archways and flying buttresses rather than simple walls. Maybe we don't need any of the specialized Legos that we think might be in that next bigger box on the store shelf; maybe we only need to learn how to stack what we've got in more sophisticated ways.

Re:Tha's goint to be the NEXT BIG THING

[12357bd]

You did not read my post. It's not only about programing computers, it's all about building new machines. Can we simulate those machines? Yes, sure, but the computational cost it's prohibitive, that's why neuronal simulations are so scarce.

Read the link about Turing papers, you'll find a very interesting bunch of ideas about 'thinking machines', not 'computers'.

Re:Tha's goint to be the NEXT BIG THING

[Yvanhoe]

Neuronal simulations are so scarce because we fail at finding a good algorithm for them to implement complex and interesting behaviors that go beyond approximating a function. We have plenty of CPU power to make neuronal simulations. Simulating a neuron with a capacitor and a transistor would be easy but we never had to make such an acceleration board because the lack of CPU power is not the problem. The problem is that our current algorithms do not scale up into something interesting.

Re:Tha's goint to be the NEXT BIG THING

[12357bd]

Algothims are great, really, but I am sorry, I don't believe in magic flowcharts. :)

Re:Tha's goint to be the NEXT BIG THING

[neumayr]

People use "correlation does not imply causation" in a wrong way? How is that possible? I've only skimmed that Wikipedia article you linked, but it seems it really does mean exactly what it says, things that correlate don't necessarily cause each other. Which seems obvious, at least given my superficial knowledge of the phrase.
What am I missing, in what way can that statement be used that is incorrect?

Re:Tha's goint to be the NEXT BIG THING

[ceoyoyo]

People frequently, perhaps most of the time, use "correlation does not imply causation" in an argument to mean "correlation does not imply a connection" which is false - correlation DOES imply a connection.

For example, there was a story on Slashdot about violent crime being linked to atmospheric concentration of lead. Naturally several someones piped up with "this research is crap, correlation does not imply causation!" The study only concluded there was a link between the two factors, which is indeed implied by a correlation.

Taking that example, finding a correlation between atmospheric lead concentration and violent crime doesn't mean that lead in the atmosphere causes violent crime. There are two other possibilities: that violent crime causes lead in the atmosphere, or that some third factor causes both lead in the atmosphere and violent crime. Any of those three possibilities is interesting.

Also, as the Wikipedia article says, finding a correlation is indeed a requirement for showing a causal relationship, it just isn't sufficient by itself. A study that shows a correlation is interesting, and a necessary step towards showing a causal relationship.

The section in the wikipedia article entitled "Usage" discusses the problems with general usage of the phrase:

In the mathematical sense, it is always correct to say "Correlation does not imply causation". However, the word "imply" in casual use loosely means suggests rather than requires. The idea that correlation and causation are connected is certainly true; correlation is needed for causation to be proved.
However, in logic, the technical use of the word "implies" means:
To be a sufficient circumstance.
This is the meaning intended by statisticians when they say causation is not certain. Indeed, p implies q has the technical meaning of logical implication: if p then q symbolized as p â' q. That is "if circumstance p is true, then q necessarily follows."
In contrast, the everyday English meaning of "imply" is:
To indicate or suggest.
Edward Tufte, in a criticism of the brevity of Microsoft PowerPoint presentations, deprecates the use of "is" to relate correlation and causation (as in "Correlation is not causation"), citing its inaccuracy as incomplete.[4] While it is not the case that correlation is causation, simply stating their nonequivalence omits information about their relationship. Tufte suggests that the shortest true statement that can be made about causality and correlation must be at least expanded to either
Empirically observed covariation is a necessary but not sufficient condition for causality.
or
Correlation is not equal to causation; it is only a requirement for it.

Re:Tha's goint to be the NEXT BIG THING

[Mandrel]

With sufficient digital computing power you can simulate any analog system, to any degree of accuracy. Deterministic, stochastic, and even non-deterministic/chaotic/quantum (though I doubt this will be required).

Digital gives you enormous flexibility, and will be the tool of choice for reverse-engineering the brain.

Re:Don't forget also

[CarpetShark]

(Without contempt or disrespect) religion is a great example of how far you can get with an incomplete model. Enlightenment, which some would argue is the highest human state, is taught with nothing more than vague contradictions that hint at a different way of thinking. Most religions use similar techniques to some extent, and I suppose most education must to some degree as well.

That said, I think religion could not have come first, as it's basically a specialised educational system. Besides, you can't teach religion before you teach words, objects, etc.

Re:Why not renaming it to memistor?

[Anne Thwacks]

Putting "mr" in a word can lead to pronunciation difficulties,

For who? Think of it as mem'ristor. There how hard is that? It is true that the pronouciation of the letter "r" is quite different in Sierra Leone and Japan, but its hardly a major problem, and the presence of the "m" in front of it isnt a problem for anyone I know.

The idea that the devices are a "major breakthrough" is a problem though - how do these differ from any amount of other devices producing "negative resistance" through phase change? As other have pointed out, its slow and awkward to use,

A viable memristor based FPGA might be interesting, and more practical than other memory applications I guess. Probably more patentable than chocolate-chip-cookies as well (but not a lot - it is clearly obvious to those "sufficiently skilled in the art" ie me.)

Not Nature...REVISITED...perfect sense...

[griffinfinity]

Interesting information...

  The only bones I have over this is what is 'artificial' in regards to intelligence? It being that mankind has been on a mission to recreate 'itself' since day 1, and everything we do is support in that respect, there is nothing artificial at all about what we seek to fashion. We are not replicators on Kirk's tugboat, we are those who seek to become that which we cannot find, the Creator. I dunno, wasn't it Water's who mumbled something about it all making perfect sense?

Re:Not Nature...REVISITED...perfect sense...

[ichigo 2.0]

Artificial usually means man-made, not unnatural.

And how exactly do you exactly plan

[msgmonkey]

to implement a proper neural network on a von neumann type architecture, it's like trying to fit a square into a circle. So the developments have been in making special processors that work closer to real neurons but still digital. Memristors allow them to get closer to the real thing. Like the article states they did n't even have the tools to test these because of their analogue nature so we're at the begining here.

The purpose here is n't to get faster hardware, a computer can add two numbers together orders of magnitude faster than a person, but try and get a computer to tell if a picture I give it is male or female or if there is even a person at all in the picture. It does n't matter how fast your hardware is your bubble sort is always going to suck vs a quick sort.

Re:And how exactly do you exactly plan

[Marble1972]

it's like trying to fit a square into a circle.

I agree - but so is running particle simulation software on today's architecture. Sure it would be better if we could have have a (parallel) processor per particle calculating all the forces impacting upon it down to planck length precision. But the point is with our limited multiprocessing power - we can achieve very good results in a more or less in a timely fashion.

My main point being - if we had the algorithm - it could be proven on today's architecture. After all - you can demonstrate a learning architecture with just matchboxes & Jelly beans and that's damned slow (but tasty!)

Take the person/picture problem and reduce the picture down to 50x50 32bit colour pixels... that's not really a lot of data to process to get a result. Training takes longer of course... but you can at least prove your algorithm within a reasonable timeframe (your PhD thesis perhaps) ;)

Re:And how exactly do you exactly plan

[raftpeople]

to implement a proper neural network on a von neumann type architecture, it's like trying to fit a square into a circle.

A feedforward neural network can be executed just fine on a von neumann architecture. At some point, we will be able to exceed human brain capacity (ignoring recurrence for now) on von neumann, it's just a matter of time.

However, can a recurrent network be properly simulated on a von neumann architecture? Not sure. The problem in that case is that multiple things are happening simultaneously that are not independent. I guess if you had enough computing power, you could have a time slice so small that things that were supposed to be simultaneous were so close to being simultaneous that you would only notice the difference after billions of years. Maybe.

Artificial intelligence?

[pieterh]

The amazing thing is that we consider individual brains to be "intelligent" when it seems pretty clear we're only intelligent as part of a social network. None of us are able to live alone, work alone, think alone. The concept of "self" is largely a deceit designed to make us more competitive, but it does not reflect reality.

So how on earth can a computer be "intelligent" until it can take part in human society, with the same motivations and incentives: collect power, knowledge, information, friends, armies, territories, children...

Artificial intelligence already exists and it's called the Internet: it's a technology that amplifies our existing collective intelligence, by letting us connect to more people, faster, cheaper, than ever before.

The idea that computers can become intelligent independently and in parallel with this real global AI is insane, and it has always been. Computers are already part of our AI.

Actually, the telegraph was already a global AI tool.

But, whatever, boys with toys...

Re:Artificial intelligence?

[dcherryholmes]

But I could stick you on a deserted island all by yourself and you would still be intelligent, right? I'm not denying that we are deeply social creatures, nor that a full definition of an organism must necessarily include a description of its environment. But I think you are confusing the process by which we become intelligent with intelligence itself.

Re:Artificial intelligence?

[hitmark]

and here i keep observing that the overall intelligence in a room drops by the square of the number of people in said room...

Re:Artificial intelligence?

[Hurricane78]

None of us are able to live alone, work alone, think alone.

Did you come up with this because of your own ability to do so?
Because except for reproduction, we can easily survive our whole life alone.
Sure it will be boring. But it works.

The idea that computers can become intelligent independently and in parallel with this real global AI is insane, and it has always been.

Says who? You, because you need it to base your arguments on it? ^^
You will see it happening in your lifetime. Wait for it.

Re:Artificial intelligence?

[Paradigma11]

Actually we usually don't consider brains intelligent, we consider humans intelligent. As for needing a society to be intelligent, well you need all kinds of stuff to be able to show intelligent behaviour (air is a biggie). I do not think that considering ourselfs as distinct parts of reality as false. This problems only originate because of western philosophies use of language and set theory in this regard. An indian philosopher once said: All problems of western philosophy exists because you are able to say that "something is" without adding an attribute like blue.

Re:Artificial intelligence?

[4D6963]

The amazing thing is that we consider individual brains to be "intelligent" when it seems pretty clear we're only intelligent as part of a social network. None of us are able to live alone, work alone, think alone. The concept of "self" is largely a deceit designed to make us more competitive, but it does not reflect reality.

No, you're completely wrong. It's sufficiently obvious why that I don't feel the need to elaborate.

Actually, the telegraph was already a global AI tool.

No, it's called a network. You seem to fail to see the difference between a network and an intelligence. I don't think you know what intelligence means.

Re:Artificial intelligence?

[mugurel]

I believe that what original poster wanted to express is that it makes no sense to talk of intelligent behavior without a context, and in the limit, a social framework. An action of an individual can be called intelligent in such a social framework, but if you take away that context, what criteria are left to judge the action as an intelligent action?

The point is that the meaning of the predicate "intelligent" is very complex. Unlike "rational", which depends only on (roughly) your definition of goals and actions.

Re:Artificial intelligence?

[DMoylan]

but we are social animals. a simple illness or injury will kill a lone human unable to feed themselves temporarily whereas a human in most societies will be cared for until they are literally back on their feet.

but that is a fully developed adult isolated on an island. a human is intelligent not solely because of their genetics but of the society they grow up in. look at the numerous cases where children have been reared by animals. http://en.wikipedia.org/wiki/Feral_child#Documented_cases

most of those have terrible difficulty adapting at a later than normal age to functioning in human societies. most have difficulty in communicating beyond the most basic needs. we learn so much in those crucial first few years. fire, wheel, levers

i would say that if you took 1000 children (pre school) and put them on that large island in such a way that most would be able to feed themselves you would not end up with intelligence for quite a while. how many generations would it take before language and tool use would be considered intelligent?

i still think turings test is the best base line for recognition of artifical intelligence. yet chatbots fool people every day. http://en.wikipedia.org/wiki/Chatterbot#Malicious_chatterbots

are they intelligent? perhaps we just need a new definition of what is artificial intelligence.

my 2c

Re:Artificial intelligence?

[dcherryholmes]

I'm not disputing anything that you wrote. I'm simply saying that the quality we loosely describe as "intelligence" can inhere in an isolated autonomous unit, even if it could never have naturally arisen that way. Therefore, I conclude that it would be possible to have AI in a set-top box. Or at least, there's nothing in our natures that refutes it.

Re:Artificial intelligence?

[cenc]

Actually yes and no. This dog has been beat and beat again fairly well in Philosophy of AI.

I appreciate what you are trying to get at, but your example is flawed. A true Islands examples for examining Intelligence in relation to AI are more like Helen Keller before she learned language or perhaps faro Children. Dropping a fully functional adult on an island, that has already learned language, culture, and essentially has learned to internalize or make self-reporting use of what is normally overt verbal expressions of internal states would just be a human on vacation by themselves. Essentially, your thoughts are not yours. They are on loan.

The "intelligence" is in fact a social construct. Yea, there is some sort of minor level intentionality involved in things like lizards, dogs, cats, and such, but fully robust use of language and thus culture really are the end game of AI. Intelligence, at least as far as we know it, is bench marked against human intelligence (thus the "artificial" part). Even relatively simple machines such as thermometers and coke machines are bench marked for their "intelligence" against the human social definition of "intelligence".

Re:Artificial intelligence?

[Chryana]

If, by your definition, the Internet is an AI, then your definition of AI is meaningless (and useless for anyone working in that field). Your post reeks of ill-deserved elitism and the message it conveys is incredibly depressing: individually the human is nothing/we already have AI, so we have nothing to reach for. I'm not going to argue about the first part, since I do not think it deserves any answer, but I'll say about the second part that we would never get true AI if most people thought like you do.

Re:Artificial intelligence?

[The End Of Days]

Ah so you wish for the rest of us to take your mental masturbation as seriously as you do. I'm gonna pass, I'm more interested in pragmatic implications than I am in stroking off.

Re:Artificial intelligence?

[cenc]

Sorry, that would be guys with their Matrix / Hal fantasies pissing millions of dollars in research money away in labs around the World, with no real coherent plan or idea of what it is they are after.

Thus, we have the nearly weekly sensational articles that claim 'we stuck our peckers in a light socket in a new way and discovered AI and God all at the same time' type articles on slahdot over and over and over again.

On a practical note however, I do suspect it will be the porn industry that ultimately makes the real breakthrough in AI research. Just look what they did for Internet.

   

Re:Artificial intelligence?

[gronofer]

I don't agree. Real progress can be made with one person obsessing over an idea. A committee would only serve to retard progress. The memresistor story is a perfect example. The concept of "self" is probably just a consequence of holding a sufficiently detailed model of reality - one that must include the self.

Re:Artificial intelligence?

[neumayr]

Because except for reproduction, we can easily survive our whole life alone.

Really? It's been tried, mostly in the 19th century - letting human children grow up in isolation, isolating grown people (can't find a link right now for either)... Anyway, as far as I can recollect, it didn't work. People go insane when isolated, children die.

You're right of course

[msgmonkey]

but on the other hand a neuron works with electrochemical signaling and the design seems to be quite good :)

Free transistors

[w0mprat]

Transistors are naturally analog, it's only that we force them to be digital. If we are prepared to accept more probabilistic outputs then there are massive gains to be had http://www.electronista.com/articles/09/02/08/rice.university.pcmos/. Work is being done with analog computing too.

I think memristors will be complimentary to existing rather than a revolution on their own yet analog transistors would have George Boole flip-flopping between orientations in his grave.

Re:Free transistors

[Twinbee]

I think even with 1000x performance, it will be hard to return to analog. There's something about the 100% copyability of data, determinism and exactness of digital which analog can't hope to achieve.

Maybe 1,000,000x would veer me over however...

Re:Why not renaming it to memistor?

[svunt]

Putting "mr" in a word can lead to pronunciation difficulties, just google for words containing "mr" then exclude all abbreviations of mister to find how rarely the sequence it's used. Renaming it to "memistor" would help greatly. Also, the wikipedia page for memristor already contains a reference to memistor.

The 'm' and 'r' are in different syllables, so it's really not an issue. I assume you can handle 'Tim Robbins' so you can handle 'memristor'

whatever

[jipn4]

In the 1970's, the big breakthrough was supposedly tunnel diodes, a simpler and smaller circuit element than the transistor. Do our gadgets now run on tunnel diodes? Doesn't look like it to me.

Re:Thanks what I meant.

[ardor]

First, we don't have glorified calculators on our desktops. Calculators usually aren't turing complete, PCs are. As for the neural networks, while there are many problems that need to be overcome, digital technology isn't one of them.

Could this explain memory loss in old age?

[abhikhurana]

If brain were indeed made of memoristors and these had finite write cycles, could it be that once we have reached these write cycles, the memoristors stop of being any use. Ofcourse the brain would try to minimise dmage to memoristors by spreading the data around but you will eventually reach a limit and eventually the same memoristors would be overwritten again and again, until eventually you start reaching the write limit for some of these, which might explain why we start losing memory after reaching 30s or so.

I suppose the way to check it, potentially, would be to see if people who have impaired senses (e.g. someone who is deaf or dumb etc.) show better brain functions in older age, as they didn't have that much data to store as someone who was getting data from all the senses.

Like we need more Artificial Intelligence...

[3seas]

... don't we have enough people producing this already?

Not really Turing complete.

[reiisi]

Effectively Turing complete within a certain range of speeds and requirements for state memory.

But the tape is finite.

So, yes, glorified calculating machines. (The boundary between is not as clearly defined as you assert.)

The first time --

[reiisi]

Adam and Eve.

Or, if you don't get the reference, us.

Humans have been doing this as far back as there have been humans. It is one of the things which sets us apart from the other animals. Or, it might be argued that this is just another way of looking at the only thing that separates us from the other animals.

enlightenment

[reiisi]

Some people believe that, in true religion, enlightenment is the realization of a rational basis to existence.

That is, half of enlightenment is the realization of the rational basis, and the other half is the realization that mortality pushes that rational basis ultimately beyond (mortal) human reach.

There seems to be some division as to whether giving up on understanding is preferred, since mortality is an absolute limit.

And there seems to be some further division as to whether mortality is really an absolute limit.

(And I see a metaphor here in the infinite tape of a Turing machine.)

Practically Turing complete.

[reiisi]

Not until we have infinite tape and infinite time to process the tape are our computers truly Turing complete.

Moore boasted that technology would always be giving us just enough more tape. I'm not so sure we should worship technology, but so far the tech has stayed a little ahead of the average need.

Anyway, this new tech may provide a way to extend the curve just a little bit further, keep our machines effectively Turing complete for the average user for another decade or so.

Or not. If Microsoft goes down, the average user may soon realize he has been seriously duped about computational needs.

Re:Practically Turing complete.

[technofix]

Human brains are finite and we have finite intelligence. Small computer based intelligences should be capable of a few Hunekers worth - something clearly detectable. But Turing completeness is a red herring. I believe intelligence is possible without Turing completeness. Our brains can't even do arithmetic; in fact, we don't have numbers in the brain and no low-level encoding mechanisms that would allow for them to be represented. Lowest levels of the brain don't use logic, arithmetic, or models. Minsky's criticism of Perceptrons fails or is irrelevant on multiple levels. 1. Slightly more advanced connectionist systems are easily Turing complete. 2. Brains are not perceptron like systems. 3. Turing completeness is not required for intelligence.

Practically Turing complete.

[reiisi]

Woops. Posted this below in the wrong sub-thread. Oh, well, post it here, too, with this mea culpa.

Not until we have infinite tape and infinite time to process the tape are our computers truly Turing complete.

Moore boasted that technology would always be giving us just enough more tape. I'm not so sure we should worship technology, but so far the tech has stayed a little ahead of the average need.

Anyway, this new tech may provide a way to extend the curve just a little bit further, keep our machines effectively Turing complete for the average user for another decade or so.

Or not. If Microsoft goes down, the average user may soon realize he has been seriously duped about computational needs.

woops

[reiisi]

Meant that in response to this.

wrong level of complexity

[reiisi]

Kludge a lot of state machines together and you can simulate stack machines to a certain limit.

Kludge a lot of context free grammars together and you can simulate a context-sensitive grammar within certain limits. But it takes infinite stack, or, rather, infinite memory to actually build a context-sensitive grammar out of a bunch of context-free grammar implementations.

Intelligence is at least at the level one step beyond -- unrestricted grammar.

(Yeah, I'm saying we seem to have infinite tape and infinite stack, even though mortality is a little hard to see beyond.)

Re:wrong level of complexity

[CarpetShark]

Intelligence is at least at the level one step beyond -- unrestricted grammar.

I don't see how you can claim unrestricted grammar when every language I know of uses the concepts of nouns, verbs, etc. Surely an unrestricted grammar would mean completely alien languages which in no way are directly translateable into other human languages.

Re:wrong level of complexity

[bstamour]

Not the grammar you're thinking of. In Computer Science grammars are used to represent languages, and the sentences that can be represented by them. For example, the language that generates all sentences of the form a^n b^n, meaning n a's followed by n b's. Grammars in this sense are a collection of rules, such as: S -> aSb We say that S is a terminal, and a and b are non-terminals. Now there are restrictions on the types of rules we can have, which creates a hierarchy of grammars, where regular grammars are on one side of the spectrum (one terminal S creates one non-terminal a), and unrestricted grammars are on the opposite side (no rules). Unrestricted grammars are obviously much more complicated to deal with, but their power is immense. Hence when the grandparent talked about using an unrestricted grammar, he meant that we don't have the power to deal with it, but if we did, hoooo boy :)

Re:wrong level of complexity

[An Onerous Coward]

I'll admit I'm a bit rusty on the subject, but I'm just not seeing how grammars (context free, context sensitive, unrestricted, whatever) have anything to do with intelligence. What's the claim actually being made there?

A quick Wikipeding says that a Turing Machine is equivalent to an unrestricted grammar, the way a finite state automaton is equivalent to a regular expression. We use Turing-equivalent machines on a daily basis (to the limits of your computer's memory), so I'm not sure what it means to say that "we don't have the power." Is there evidence that the calculations that would lead to true intelligence would require unimaginable amounts of RAM? I doubt there is.

Or is he saying that we really haven't mastered the art of creating unrestricted grammars?

It was 1960s, and they were quickly obsoleted

[Kupfernigk]

The Esaki (tunnel) diode is a two terminal device which basically exists in two states (I am simplifying, I know) at two different currents. Its weakness is that (a) it requires a current source to keep it in one or the other state and (b) both input (changing state) and output need amplifying devices. As soon as cmos become fast enough things like tunnel diodes were dead in the water because a cmos transistor does its own amplifying, and requires almost no power to keep in one state rather than the other.

Therefore, a device which requires effectively no power to keep in one of two states, and has much greater speed than either flash or magnetic domains would be a step forward compared to the current state of the art.

Re:Don't forget also

[CarpetShark]

Depends on your definition of teaching. Most of the education world would include being a role model, providing examples, as a type of teaching too.

Re:Why not renaming it to memistor?

[Twinbee]

Perhaps he thought it was pronounced "me mristor", as in: "Oill get me mristor out torday!".

The brain is not a computer.

[FiloEleven]

Citation.

See especially points
6 - No hardware/software distinction can be made with respect to the brain or mind,
7 - Synapses are far more complex than electrical logic gates,
10 - Brains have bodies,
and the bonus - The brain is much, much bigger than any [current] computer.

It's past time for this idea to die.

Re:The brain is not a computer.

[Anonymous Coward]

Citation.

See especially points

6 - No hardware/software distinction can be made with respect to the brain or mind,

Software can be expressed as hardware. For instance, decoder chips exist for video that take the load off of a processor.

7 - Synapses are far more complex than electrical logic gates,

Fine, so they are. Model them with multiple gates, or something else.

10 - Brains have bodies,

Computers have peripherals.

and the bonus - The brain is much, much bigger than any [current] computer.

and today's computers are much, much bigger (transistor count) than computers of a few decades ago, but both are still computers.

It's past time for this idea to die.

It shouldn't. Computers manipulate data, which brains do very well. It's a worthy goal to try and emulate it.

Re:The brain is not a computer.

[FiloEleven]

You clearly did no further reading other than the topic headings I provided. I should expect no more from an AC, yet somehow the optimist in me keeps wanting to believe that people are seeking truth instead of trying to reinforce what they already know.

If you (or anyone else) can raise reasonable objections to what's in that article, I'm willing to explore the topic. Otherwise, don't waste my time.

Re:The brain is not a computer.

[ignavus]

Difference #1: brains are analogue; digital computers are digital; analogue computers are analogue ... um, what was his point again? Computers are no more digital than human beings are males. That is, some are and some aren't.

Difference # 10 "brain have bodies" - um, my body has a brain. My brain is a body part.

Could not be bothered reading the rest.

Re:The brain is not a computer.

[An Onerous Coward]

No, he made some excellent points, which you decided to ignore.

The article makes for fascinating reading, but it isn't at all clear what you're trying to accomplish by citing it. The brain/computer analogy is a useful one for dealing with certain issues. It has its limits. Ultimately, computers today are based on different principles than the human brains.

The article is just a warning (seemingly directed towards cognitive scientists) to take care in drawing conclusions about how the brain works from their understanding of computers. What are you trying to claim? That artificial intelligence is impossible? That it cannot be simulated on a Von Neumann style machine, no matter how fast a processor or how big a RAM store it has? Or maybe that only brain-style AI is possible? I don't think any of those things are true. They certainly can't be derived from the evidence presented.

Regarding specific points:

1) It is entirely possible to build analogue circuits.

2) A content-addressable system can be built atop a standard memory system. I don't see a theoretical problem there.

3) Any parallel computation can be performed in a serial way.

4) There are processors that have no system clocks. See "asynchronous computing."

5) That's not to say that human-style memory couldn't be simulated within a computer-style system.

6) While the point is good, it's possible to erase that distinction on computer hardware as well. You can take essentially any set of software
rules, map the rules to a set of transistors and pathways, and etch those onto a chip. As simple as that, you have a chip that has this very feature: there is no distinction between hardware and software.

7) It's not clear which of the differences are vital and which have minimal effect on the overall process. But all the differences could be embodied in a sufficiently complex neuron simulation.

8) Still, software that behaves that way could be executed inside your average laptop. He's pointing out differences, not fundamental barriers to artificial intelligence.

9) Again, not part of the standard computer architecture. But software living atop it could have such behavior.

10) If anything, this point says that we're dumber than we thought, because we let our environment remember things for us. That would, if anything, make it easier for machines to outperform humans.

Bonus) That state of affairs may not last forever, and it's not clear which items would have to be simulated in order to make a brain work in software. Maybe you have to simulate every glial cell. Maybe an entire class of neurotransmitters could be removed entirely from the simulation, or replaced by a simple adjustment of its propensity to fire.

New Scientist screwed up on this one. As far as I can glean from the article, memristors are magic fairy dust that could make artificial brains work. No real explanation as to why, beyond "We're New Scientist, and we'd sell our own mothers for a few more page clicks."

Re:The brain is not a computer.

[FiloEleven]

No, he made some excellent points, which you decided to ignore.

I disagree. The responses were clearly knee-jerk reactions to the summaries I quoted. "Computers have peripherals?" Please. That response is either disingenuous or shows a lack of understanding of the tight integration of our massive sensory apparatus and brain. The difference between that and a peripheral, even something as complex as, say, the entire nav system for a Mars rover, is orders of magnitude apart. There is simply no man-made sensory network anywhere near the complexity of the human body, or even that of most animals.

The article makes for fascinating reading, but it isn't at all clear what you're trying to accomplish by citing it. The brain/computer analogy is a useful one for dealing with certain issues. It has its limits. Ultimately, computers today are based on different principles than the human brains.

The article is just a warning (seemingly directed towards cognitive scientists) to take care in drawing conclusions about how the brain works from their understanding of computers. What are you trying to claim? That artificial intelligence is impossible? That it cannot be simulated on a Von Neumann style machine, no matter how fast a processor or how big a RAM store it has? Or maybe that only brain-style AI is possible? I don't think any of those things are true. They certainly can't be derived from the evidence presented.

I claim that neural networks alone will not get you consciousness. I think it highly unlikely that consciousness can be achieved on Von Neumann architecture, yes, and impossible for human consciousness to exist on silicon. I'm not sure what you mean by "brain-style AI."

I didn't make it very clear that this was my position in my original post. If you're not concerned with conscious AI, then skip the rest because that's mainly what I am arguing against.

Regarding specific points:

1) It is entirely possible to build analogue circuits.

Agreed.

2) A content-addressable system can be built atop a standard memory system. I don't see a theoretical problem there.

Expanding on the article, will this content-addressable system experience decay and interference? One of the ways we learn is by replaying important things in our minds and strengthening those connections. If that holds for very young minds (which we have no reason to doubt) then it may be that without interference and decay, we could never begin to make sense of the world. If that is the case, an AI would need some similar way of filtering input.

3) Any parallel computation can be performed in a serial way.

I'm skeptical. This certainly holds for digital computers, but I don't think it does for analog. Someone who knows more about analog computers and circuits can probably shed some light here.

4) There are processors that have no system clocks. See "asynchronous computing."

Asynchronous computing is not a similar mechanism to what the brain does. Asynchronous is still sequential, and there's even a clock, it's just that its rate isn't limited by worst-case instruction calls. FTA:

"The speed of neural information processing is subject to a variety of constraints, including the time for electrochemical signals to traverse axons and dendrites, axonal myelination, the diffusion time of neurotransmitters across the synaptic cleft, differences in synaptic efficacy, the coherence of neural firing, the current availability of neurotransmitters, and the prior history of neuronal firing."

In other words, you're going to have to run a physics simulation if you want the brain's kind of clockless processing.

5) That's not to say that human-style memory couldn't be simulated within a computer-style system.

Agreed, this is almost

New brain cells are created through life

[Kupfernigk]

The old idea that brain cells are lost and new ones don't form has turned out to be just plain wrong. The evidence is that the more you think about a subject, the bigger relevant parts of the brain get. For instance, London cab drivers have to memorise large chunks of the road system to pass a test, and it has been shown that the relevant part of the brain does actually grow during the process.

Constructionist vs. Instructionist

[technofix]

Yes. That's how we learn 99+ % of our words. And most of the grammar. And all of how to interpret visual stimuli. And our motor control. And 99.99+ % of the world knowledge we finally end up with. Learning (by pull) is Constructionist. Getting taught (by push) is Instructionist. The latter is severely overvalued; the former is what Montessori and other successful schools use. We used to think "The student is an empty vessel to be filled with knowledge". Today we can blame most of the problems with our public school system on the refusal to drop this misconception.

Re:Not nature...

[technofix]

The bible is a very heavily peer reviewed document. It is an excellent example of the problems with peer review as a proxy for veracity.

Just make each CPU a usb stick for $5.

[cheekyboy]

if each core2 or atom cpu could be placed in a small USB stick for $5, then one can build the biggest usb chain with 1000s of $5 usb cpu sticks.

What we need is any easy was as easy as usb sticks to increase cpu horse power of desktops, without the need for cumversome sockets+fans+sinks. Maybe if each USb stick contained 256 x Z80 cpus, each running at 500mhz, at 500ma, then we could easily allow the freedom to grow. And if we could make multi layer cpus, ie say 100 layers of 256 cpus, that would be some advance. Perhaps we should strive for more quantity of simpler CPUs at optimum mhz speeds. The old 68k cpu was powerful even at 7mhz. Now that same design cpu can be made to run easily at low voltage and at 500mhz easily, but rather than 1, we could have 256 of them, which would be no more complex at 45nm than an atom cpu. GPU style.