Vinge on the singularity

quote:

---

Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended.

Is such progress avoidable? If not to be avoided, can events be guided so that we may survive? These questions are investigated. Some possible answers (and some further dangers) are presented.

---

quote:

---

What is The Singularity?

The acceleration of technological progress has been the central feature of this century. I argue in this paper that we are on the edge of change comparable to the rise of human life on Earth. The precise cause of this change is the imminent creation by technology of entities with greater than human intelligence. There are several means by which science may achieve this breakthrough (and this is another reason for having confidence that the event will occur):
There may be developed computers that are "awake" and superhumanly intelligent. (To date, there has been much controversy as to whether we can create human equivalence in a machine. But if the answer is "yes, we can", then there is little doubt that beings more intelligent can be constructed shortly thereafter.)

Large computer networks (and their associated users) may "wake up" as a superhumanly intelligent entity.
Computer/human interfaces may become so intimate that users may reasonably be considered superhumanly intelligent.
Biological science may provide means to improve natural human intellect.
The first three possibilities depend in large part on improvements in computer hardware. Progress in computer hardware has followed an amazingly steady curve in the last few decades [17].

Based largely on this trend, I believe that the creation of greater than human intelligence will occur during the next thirty years. (Charles Platt [20] has pointed out that AI enthusiasts have been making claims like this for the last thirty years. Just so I'm not guilty of a relative-time ambiguity, let me more specific: I'll be surprised if this event occurs before 2005 or after 2030.)

What are the consequences of this event? When greater-than-human intelligence drives progress, that progress will be much more rapid. In fact, there seems no reason why progress itself would not involve the creation of still more intelligent entities -- on a still-shorter time scale. The best analogy that I see is with the evolutionary past: Animals can adapt to problems and make inventions, but often no faster than natural selection can do its work -- the world acts as its own simulator in the case of natural selection. We humans have the ability to internalize the world and conduct "what if's" in our heads; we can solve many problems thousands of times faster than natural selection. Now, by creating the means to execute those simulations at much higher speeds, we are entering a regime as radically different from our human past as we humans are from the lower animals.

---

I have a hard time finding flaws in his logic. Maybe the time-frame is overly foreshortened.

quote:

---

Computer/human interfaces may become so intimate that users may reasonably be considered superhumanly intelligent.

---

While I agree the time frame might be a little compressed, I think we're on track.

It dawned on me when they put an electrode, first into chimpanzees, then later, humans, which enabled them to move a cursor on a computer terminal just with their brains.

While we look at this and other technologies as computer prosthetics, surely applications as enhancements are only a millisecond behind.

I envision having the net, or at least your "favorites" hard wired into your brain.

As far as designing super-human beings goes, if we do indeed develop the technology to expand the human life span to the point where the sum of individual knowledge no longer goes down the crapper when we die or fall prey to senile dementia, we might explore the limits of the human mental capacity. We may even enhance it. But there are limits to the biological hardware that we are as yet completely ignorant of. Time will tell, of course.

AI research has been around for quite a while, but Bioinformatics is relatively new. And that's where you'll get the technology for human-digital interface. It's a long way away from the kind of sophistication required to significantly enhance human intelligence.

I think Vinge and others who are interested in a singularity that would create a self-aware machine or a super-intelligent human being have been reading too much Asimov. It's an interesting vision of the future, but really too full of "ifs" and "whens" to be a technological course plotted with a foreseeable outcome.

[ 14 May 2003: Message edited by: Rebecca West ]

quote:

---

Large computer networks (and their associated users) may "wake up" as a superhumanly intelligent entity.

---

quote:

---

It dawned on me when they put an electrode, first into chimpanzees ...which enabled them to move a cursor on a computer terminal just with their brains.

---

So now every computer will come equipped with a monkey! (Albino monkeys for Imacs)

quote:

---

And I doubt computer networks will suddenly "wake up" either, at least not in the sense we understand. There is no body, no driving need to form some sort of goal (what, are they gonna wake up one day and think that the subjugation of humans is a great thing and enslave us all? Why would it wake up to that? Do machines care about money and power? Would a sufficiently complex computer even be able to grasp that?).

---

With quantum computing, and a host of other technological innovations, the hardware limitations aren't an issue anymore. But still, even those who're involved in the most sophisticated AI research and development (in the private corporate sector, natch) cannot claim to have produced a self-aware machine yet.

Edited to add: oh, okay. That makes sense. If I'm understanding it correctly, I imagine that it would be some sort of self-organizing criticality, that would eventually transition to a self-aware "state" for the machine.

[ 15 May 2003: Message edited by: Rebecca West ]

But I want to go back to Mando's link..

quote:

---

Originally posted by clockwork:

And I doubt computer networks will suddenly "wake up" either, at least not in the sense we understand. There is no body, no driving need to form some sort of goal (what, are they gonna wake up one day and think that the subjugation of humans is a great thing and enslave us all? Why would it wake up to that? Do machines care about money and power? Would a sufficiently complex computer even be able to grasp that?).

---

It is possible, and I don't think in the immediate future, to expand human capabilities with a computer chip. Intelligence is another matter. For example, we could put an entire medical encyclopedia onto a chip and implant it into a human with a direct interface into the brain allowing immediate acess.

I think, "Oh no, these are all the signs of a heart attack!"

So I grab my knife and some PVC tubing that I keep around the house, knowing that with my trusty chip in the back of my head, I know what to do. I must do an incision here below the right ventricle. With luck, I'll insert the tubing and complete a bypass.

I make the first cut… "Ewwwwww! Blood!"

I don't think there is an adequate understanding of what "human intelligence" is to ascertain "super human" relative to.

Even if we have an interface that allows us to send instructions to a machine just by "thinking", or with minimal abstraction between squishy neurological synapse firing and calculation, how would this be different from controlling a hammer, a car or a fork with your mind?

To borrow from Arthur C Clarke, it seems that computers are sufficiently advanced that most people can't distinguish their operation from magic. Just because most people don't understand computers doesn't make them (computers) intelligent, despite what some people think a Turing test ascertains.

The assumption inherant in most models of AI is that there is a consistent computational model of conciousness just waiting to be inevitably uncovered. That is to say, that our entire conciousness can be reduced and accurately modelled as a set of binary operations.

That really smart people, (including some cognitive scientists) believe in a strictly computational model of mind, shows more about the limitations of our ability to precieve things than it does about the possibility of replicating it.

Edited to add: do I smell Searle in the air? I hope not. How do you know Searle's head is not a Chinese Room?

[ 21 May 2003: Message edited by: Mandos ]

quote:

---

However, I will make a note to re-read it and in light of comments here, edit this post.

---

There are only about 3 or 4 people that certain threads atract intelligent comments... I hope I'm one of them, nut I never usually am: You may say that here, but you said that in my thread too.

I think these Arthur Clarke quotes are more cogent to the discussion:

1. When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.

2. The only way of discovering the limits of the possible is to venture a little way past them into the impossible.

In other words, it's usually a mugs game to ascribe limits to what is technologically feasible.

[ 21 May 2003: Message edited by: JimmyBrogan ]

I like his reference to the hammer.

But this is what it comes down to isn't it? Are computers intelligent because they can choose from a number of preprogrammed options based on a number of selected scenarios?

The real issue is can think the impossible?

Anyway, yes, I am saying that faith in a strictly computational model of mind is ridiculous, precisely because the very very best you could hope for is that it is internally consistent.

Faith in any model of determinism seems silly, as if it was a consistent deterministic model that was externally consistent with the rest of the universe, you woudln't need faith would you?

A tautology or solipsism, maybe, but an important one nevertheless.

As far as technologically feasable goes, what is to say that our machines and networks aren't intelligent now? I can see how someone might have problems with notions of subjectivity (relativism in general), but it becomes a very important factor when you are talking about who decides what is or isn't intelligent.

Further, which is intelligent? The software? The hardware? Some kind of magic gestalt of one operating upon the other? I am all for gestalts, but I wonder if such a relationship would also apply to musical notation and the instrument it is played upon.

I would be willing to say that listening to Glen Gould do the Goldberg variations is communing with a disembodied but sentient "intelligence" but I'm not sure it would be consistent with any accepted deterministic scientific models of mind.

I think that conciousness falls into the catagory of things that as soon as you bound their definition within a formal system, you find yourself being right, but about the wrong
thing.

God is neither an engineer or a mechanic. There
was a great comment from one of the astronomers from the Vatican about how God is Love, and therefore the notion of scientific discoveries undermining the possibility of Gods existence was silly. Finding the edges of the universe and the smallest quanta of matter won't harm humanities experience of Love (and other things) any more than writing software could.

We can express pretty much anything formally, but I suspect that the more precisely and formally we express it, the less important/valuable/externally
consistent/true/ it is.

This is a running aesthetic theme with Godel, Hiesenberg, Shannon, Mandelbrott, Bach and Escher, among others. Maybe the reason we appreciate their work so much is because by stretching the bounds of our comprehension of their fields, they also showed us the negative space created by the ultimate limitations of their mode of inquiry.

It's a question of the effect of emphasis on our perspective, which is pretty interesting.

There is a cool book called The Philisophical Computer I half-read a while ago that deals with alot of this stuff.

(for the record, I am not an AI researcher (IANAAIR?), a statistician, or educated enough to engage anyone on the subtleties of durkhiem vs derrida vs minsky vs keanu reeves. I'm really just a hacker working a night shift trying to stay out of trouble.)

[ 22 May 2003: Message edited by: batz ]

[ 22 May 2003: Message edited by: batz ]

quote:

---

Anyway, yes, I am saying that faith in a strictly computational model of mind is ridiculous, precisely because the very very best you could hope for is that it is internally consistent.

Faith in any model of determinism seems silly, as if it was a consistent deterministic model that was externally consistent with the rest of the universe, you woudln't need faith would you?

---

In any case, your claim about a "consistent deterministic model" vs. "faith" makes no sense to me. Just because a model is deterministic doesn't mean I know it.

Is the mind deterministic? Probably not--entirely. There are very likely nondeterministic choice points that would be satisfied by some form of randomness. However, can the mind be convincingly represented in some formal language? That's a completely different question.

quote:

---

As far as technologically feasable goes, what is to say that our machines and networks aren't intelligent now? I can see how someone might have problems with notions of subjectivity (relativism in general), but it becomes a very important factor when you are talking about who decides what is or isn't intelligent.

---

quote:

---

Further, which is intelligent? The software? The hardware? Some kind of magic gestalt of one operating upon the other? I am all for gestalts, but I wonder if such a relationship would also apply to musical notation and the instrument it is played upon.

---

quote:

---

I would be willing to say that listening to Glen Gould do the Goldberg variations is communing with a disembodied but sentient "intelligence" but I'm not sure it would be consistent with any accepted deterministic scientific models of mind.

---

[ 22 May 2003: Message edited by: Mandos ]

quote:

---

Originally posted by Mandos:
The problem is, are we intelligent because we choose from a set of preprogrammed scenarios? Creativity, alas, does not come from nowhere. I do not believe it is unconstrained. If it is constrained, then there is a heuristic. It is only one more step to "can be expressed in formal terms."

---

From a link: "Instead, Searle argues that the relation between consciousness and its causal brain processes involves a kind of non-event causation such as would explain the fact that gravity (a non-event) causes an object to exert pressure on an underlying surface. Searle has put the point another way by describing consciousness as an emergent property of brain processes in the same sense that water's liquidity is an emergent property of the behavior of H2O molecules."

That seems useful. He had another principle about conciousness being "irreducable" which sounds pretty good.

The deterministic'ness of a system is relative anyway, similarly to the way that randomness is is best measured by its sufficiency. A good example of how the behaviour of seemingly soulfull things like people is bounded deterministicly, would be in how cryptographers say that "people are a poor source of entropy".

Writing an algorythm that generates suffieciently random data may qualify as super-humanly random, but it is probably reasonable to assume that it isn't intelligent, even if it develops emergent properties.

I am willing to admit to adding arbitrary restrictions on the definition of intelligence, but only because I see computationalism as requiring equally arbitrary restrictions of a different sort to be consistent.

I don't think I have fallen into the "location" fallacy, though I would speculate that computationalism may substitute "when" with
"where" and say it is free of this fallacy.

As for visualizing something in 4 dimensions, isn't that really just showing how it changes over time? 5-D is when you really have to start visually compressing things, which is
interesting when posited against Searles
notion of irreducabililty. Algebraicly we can express N-dimensions, but the jury is still out on whether expressing them makes them "real". This handily complements our discussion here, in that expressing intelligence and being intelligent are probably different things.

I thought a few years ago of researching SI,
which is Superficial Intelligence. Things that seem intelligent but really aren't. It was going to involve going to Future Bakery and asking people about these very things and measuring the length of their soliliquies. Takes one to know one, I suppose.

The argument from here generally goes on to "well do you have anything better?" to which I reply, "That isn't my burden, it's just worth noting that AI could learn a bit from some other cultural critical discourses, as they could provide some perspective on metrics for success in the field."

quote:

---

From a link: "Instead, Searle argues that the relation between consciousness and its causal brain processes involves a kind of non-event causation such as would explain the fact that gravity (a non-event) causes an object to exert pressure on an underlying surface. Searle has put the point another way by describing consciousness as an emergent property of brain processes in the same sense that water's liquidity is an emergent property of the behavior of H2O molecules."

---

Describing something as an emergent property does not itself indicate that the property does not have a reality of its own, or that there is only one way that it can emerge. But this is not the whole of Searle's argument.

Searle has, apparently, changed his mind about AI as technology has progressed. However, as I understand it, he believes that it will be achieved by copying the meat. (Hey, is Sisyphus reading this?) I, however, think that the emergent properties of the mind have a reality of their own and can be modeled in another context. Perhaps the correct position is somewhere in between the two, but this is an empirical question.

quote:

---

Writing an algorythm that generates suffieciently random data may qualify as super-humanly random, but it is probably reasonable to assume that it isn't intelligent, even if it develops emergent properties.

I am willing to admit to adding arbitrary restrictions on the definition of intelligence, but only because I see computationalism as requiring equally arbitrary restrictions of a different sort to be consistent.

---

quote:

---

I don't think I have fallen into the "location" fallacy, though I would speculate that computationalism may substitute "when" with "where" and say it is free of this fallacy.

---

quote:

---

As for visualizing something in 4 dimensions, isn't that really just showing how it changes over time? 5-D is when you really have to start visually compressing things, which is
interesting when posited against Searles
notion of irreducabililty. Algebraicly we can express N-dimensions, but the jury is still out on whether expressing them makes them "real". This handily complements our discussion here, in that expressing intelligence and being intelligent are probably different things.

---

I find that the idea that something with a "real-world" effect is "irreducible" is a destructive proposition. What are we supposed to do with that claim? It smells obscurantist to me.

Your last claim is once again an effective restatement of the "fallacy of location." How is it that "expressing intelligence" and "being intelligent" are two different things? The only way you could claim that is if you have located intelligence, and claimed a priori that it can arise nowhere else. So what happens if an alien comes up and talks to you...how do you determine whether it is "expressing intelligence" or whether it is "being intelligent"? Heck, how do I know that about you? I can just claim that you are "expressing intelligence."

I saw an amusing illustration of this point a few months ago on USENET. Suppose a pharmaceutical company came up with an extra-strength hospital painkiller with the following warning:

quote:

---

After taking the recommended dosage, the patient will likely continue to complain of extreme discomfort. Other tests may also confirm the apparent distress.
The medical staff should consider this only to be apparent pain: there can be no real pain after taking this medication. All expressions and indications of pain are actually side-effects of the painkilling agents in this medication, and not to be confused with actual pain.

---

quote:

---

The argument from here generally goes on to "well do you have anything better?" to which I reply, "That isn't my burden, it's just worth noting that AI could learn a bit from some other cultural critical discourses, as they could provide some perspective on metrics for success in the field."

My comments weren't _really_ challenging the internal consistency of the theories of AI (despite the validity of such challenges), they were to pose the question of why AI theories aren't externally consistent with some other critical perspectives. Not that they have to be, but it would be interesting to know why they aren't.

---

For example, I used to discuss biological determinism and sexual behaviour in cultural contexts here quite a bit, but the perspectives on the issue were so incompatible and the rejection was so passionate that I've mostly decided that there is no point in trying. I think the underlying problem was I was trying to dissect culture as a natural object, and people who study culture, literature, and so on have no inclination to accept that kind of analysis. How this creates a difficulty in reconciling classical AI theories with theories of culture should be obvious.

[ 23 May 2003: Message edited by: Mandos ]

[ 23 May 2003: Message edited by: Mandos ]

By the way, it is perfectly OK to private message me and tell me if you think I'm being a snot, you know.

Ok. My AI research knowledge is a bit fuzzy. However, having said that, I think Vinge's over-the-top-ness comes from his assumption of an exponential increase in machine computing power, without limit.

There are, however, practical limits (there is of course no theoretical limit save that fixed by the Uncertainty Principle). One of them is how small you can make computer chips. The other is how big you can make a computer. I think this will tend to delay the Singularity (to use Vinge's term) or to spread it out over a more manageable time frame than his conception of an ever-compressing interval between increments of machine intelligence improvements.

Where I stand on AI is that it will, some day, be possible to design truly intelligent and self-aware robots. I have a somewhat selfish reason for advocating this as well as an altruistic one: I want a robot to do all my scut work for me. I also want robots to free humans from the yoke of physical labor.

Alvin Toffler has also written of the acceleration of things that happen in societies, but where I differ from Toffler and Vinge is that I don't think humans will accept an ever-accelerating pace of change. We are psychologically resistant to some forces of change; one example is the continuing lack of desire to pay on a piecewise basis for things that people used to pay for up front.

quote:

---

Instead, in the early '00s we would find our hardware performance curves beginning to level off -- this because of our inability to automate the design work needed to support further hardware improvements. We'd end up with some _very_ powerful hardware, but without the ability to push it further.

---

Back to Vinge.

In some ways this scenario he outlines is happening.

People who use personal computers have reached a kind of consensus that they don't really "need" a better than 2.5 GHz computer. They feel they don't "need" more than 512 megs of RAM. Et cetera.

The primary driver of computing power increases among the general population (which, I assume, makes up part of the Vingean requirement for the Singularity to form) is whether or not people need to put the brute force of extra computing speed behind something. If they don't, they don't.

Sidebar on cultural analysis, though someone more schooled than me in the social sciences can tell me the linkage between that and singularities.

Quoth Mandos:

You'd think that one side would back down, incorporate or co-opt the other. The alleged militancy on the part of critical theorists against scientistic analysis is really just a product of the same ignorance that makes many engineers objectivist Randroids.

The solution is to attempt to reconcile the external inconsistencies, and assess whether they are indicative of internal ones. An example would be how any good scientist will tell you that they are in the business of collecting evidence, and that Truth is for zealots. Similarly, feminism has railed against the poor science behind things like assumptions of a biologically determined patriarchy in Nature.

So there appears to be a difficulty reconciling certain philosophical assumptions at a fundamental level. Classical AI generally assumes logic to be a natural and not a "cultural" object. How can a reconciliation take place without divesting these fundamental assumptions?

I am not claiming that the humanities and the social sciences are useless for the study of AI. Quite the contrary. The various traditions of linguistics have a lot to say--because many of them start on a mathematical basis. But a lot of other areas, well, don't. For the reasons I've mentioned.

quote:

---

The solution is to attempt to reconcile the external inconsistencies, and assess whether they are indicative of internal ones. An example would be how any good scientist will tell you that they are in the business of collecting evidence, and that Truth is for zealots.

---

But I do realize that you likely mean some kind of overarching claim of absolute knowledge. It is true that the natural sciences cannot afford this. But those of us retrograde rationalists who approach these questions from a formal and mathematical perspective must be able to make Truth claims. And natural scientists too must have some kind of absolute epistemological basis.

quote:

---

Similarly, feminism has railed against the poor science behind things like assumptions of a biologically determined patriarchy in Nature.

---

quote:

---

To date, computers have used the binary bit -- represented by either a one or zero -- as their fundamental unit of information. In a quantum computer, the fundamental unit is a quantum bit, or qubit. Because qubits can have more than two states, calculations that would take a supercomputer years to finish will take a quantum computer mere seconds.

Due to the complexities of quantum dynamics, electrons can serve as qubits. They can exist in "up" and "down" states -- single points that are analogous to the ones and zeroes in classical computers -- or in "superposition" states, which are not single points but patterns of probability that exist in several places at once.

---

quote:

---

Theoretical physicists at Stanford and the University of Tokyo think they've found a way to solve the dissipation problem by manipulating a neglected property of the electron - its ''spin,'' or orientation, typically described by its quantum state as ''up'' or ''down.'' They report their findings in the Aug. 7 issue of Science Express, an online version of Science magazine. Electronics relies on Ohm's Law, which says application of a voltage to many materials results in the creation of a current. That's because electrons transmit their charge through the materials. But Ohm's Law also describes the inevitable conversion of electric energy into heat when electrons encounter resistance as they pass through materials.

''We have discovered the equivalent of a new 'Ohm's Law' for spintronics - the emerging science of manipulating the spin of electrons for useful purposes,'' says Shoucheng Zhang, a physics professor at Stanford. Professor Naoto Nagaosa of the University of Tokyo and his research assistant, Shuichi Murakami, are Zhang's co-authors. ''Unlike the Ohm's Law for electronics, the new 'Ohm's Law' that we've discovered says that the spin of the electron can be transported without any loss of energy, or dissipation. Furthermore, this effect occurs at room temperature in materials already widely used in the semiconductor industry, such as gallium arsenide. That's important because it could enable a new generation of computing devices.''

---

This seems like a bit of a breakthrough. They're taking advantage of a wonderful natural freebie.

[ 11 August 2003: Message edited by: JimmyBrogan ]

I think time-frame and path to a 'singularity' type event are the only unknowns. A continued evolution of 'intelligence' or self-organization will occur.

Chemical soups self organized into organic molecular structures to cells to multicellular to organisms. Complex, unfathomable but apparently inevitable.

We are here in all our self aware, complex multi-million cell structural splendor. Billions of years ago there was no life on Earth. 250 million years ago most of the life that was present went extinct. 80 millions years ago that happened again. It has probably happened many other times to various degrees of 'setback'. Each time, however, living systems relentlessly continued evolving in to more complex organizations.

To argue whether Moore's law or other potential technological bottleneck or barrier will cap this evolution is moot.

Argument about the fine-points of evolution of non-biological entities, assisted by our own creations, is moot in the big picture too. Did a philosophical Australopithecus afarensis hominid alive 4 million years ago, ponder the horror of some super-human superseding her and her kind? Maybe. But we have some of her in us as will whatever we create. Maybe we can even bootstrap ourselves into that future .. by morphing ourselves into a computational substrate. Bring it on.

It may well be that those without access to the technology die off. It may be that access to the technology becomes easier and easier. who can say.

The truly awesome aspect to this though, is that as biotech and infotech allow individuals to expand their capabilities in terms of life span and mental (storage, computational, etc) capacities, it is very likely they will dominate in one way or another. This will be similar to centres of wealth today eg the 450 odd billionaires but with the potential to have dramatically greater wealth and influence. Whether it will be benign, enlightened or otherwise remains to be seen. I think one indication that it may be somewhat enlightened or simply benign is the propensity of the world’s wealthy to philanthropy. They may choose to float all the boats, while keeping its boat riding higher and higher.

quote:

---

“In the long term, the benefits from this effort are expected to include augmenting human effectiveness and embedding these cognitive models into systems like robots and vehicles for better human-hardware interactions,” says John Wagner, manager of Sandia’s Computational Initiatives Department. “We expect to be able to model, simulate and analyze humans and societies of humans for Department of Energy, military and national security applications.”

---

New York Times, August 12, 2003

Genetic medicine is making enormous strides, and it may hold the promise of eventually making us something closer to immortal.

"Our life expectancy will be in the region of 5,000 years" in rich countries in the year 2100, predicts Aubrey de Grey, a scholar at Cambridge University.

Barrow and Tipler's argument is interesting. I am take it as a weak tea though. There are many people who theorize given sufficient computational power, it is possible to recreate the universe. The part about escaping the end of the universe by merging with all intelligence and being a deity is more sci-fi. But heck why not, eh!

quote:

---

Salk Researcher Provides New View on How the Brain Functions:

As the brain has evolved over millions of years, according to Sejnowski, it has become amazingly efficient and powerful. He says that nature has "optimized the structure and function of cortical networks with design principles similar to those used in electronic networks." To illustrate the brain's tremendous capacity, Sejnowski and Laughlin state that the potential bandwidth of all of the neurons in the human cortex is "comparable to the total world backbone capacity of the Internet in 2002."

But they point out that simply comparing the brain to the digital computers of today does not adequately describe the way it functions and makes computations. The brain, according to Sejnowski, has more of the hallmarks of an "energy efficient hybrid device."

"These hybrids offer the ability of analog devices to perform arithmetic functions such as division directly and economically, combined with the ability of digital devices to resist noise," he writes in Science.

"This is an important era in our understanding of the brain," according to Sejnowski. "We are moving toward uncovering some of the fundamental principles related to how neurons in the brain communicate. There is a tremendous amount of information distributed throughout the far-flung regions of the brain. Where does it come from? Where does it go? And how does the brain deal with all of this information?

---

I bet Rasmus has read it. Probably has a like new first edition sitting on the shelf.

Mr Raven?

A word on Mr Gheorghiu's book, if you please.

[ 08 October 2003: Message edited by: flotsom ]