Talk:Singularity/Archive 1

Comment by Jimbo Wales: I think that this is likely to happen within my own natural lifespan. I'm 34 now, and so in 2031, I will be 54. If I take care of myself, I should be able to live to 74-104, even given today's technology.

It strikes me as a virtual certainty that we will have massively cheap machine intelligence by then. And that's going to be, for better or worse, the most amazing thing that has ever happened on Earth.

While we're firing off personal opinions. It's an open question exactly how powerful a computer one would need to support consciousness (heck, that possibility is still somewhat open). A friend of mine, though, suggested this possibility: you need basically as many connections as an ape brain has. In our vacuum datarum I would go with this, since it's obvious and fits with consciousness arising evolutionarily when it did.

In that case, we have a long ways to go, since most of our computers are about as powerful as worms. Good at different things, of course, but I don't see them helping too much. So don't wait up...well, I guess you should if you can, but that's a different matter. :)

Btw, never extrapolate trends too far! Technology can't surpass physical limits, which strongly seem to exist. So its growth should be logistic rather than exponential, and we're far enough away from the limits we know of that it makes sense we can't yet tell the two apart. -- JoshuaGrosse

Oh, I agree with what you say. This is just a fun idea more than anything else.

There is a fairly respectable argument that the processing power of the human brain is something on the order of 100 million million to 1 billion million instructions per second. IBM is currently planning to have completed by 2005, a computer which will handle 1 billion million instructions per second. Of course, such an expensive ($100 million) computer will not be used for something as silly as approximating human intelligence -- it will be used to work on problems in protein folding, as I understand it. But, let's suppose that 20 years from now, a computer that powerful is cheaply available.

Given the history of computers so far, and given the technological advances that seem to be "in the pipeline", it doesn't seem totally outrageous to suggest 2021 as a date for that.

But, tack on an extra 30 years to be safe. So, will we have it by 2051? --Jimbo Wales

Oh, when I say processing power I'm not referring to speed. A fast calculator is still a calculator. I'm referring to the number of internal connections the system has. Our brains have a lot of neurons connected in very complex ways, so consciousness can emerge relatively easily. Present computers have relatively few gates hooked up to each other simplistically, so there's no such room. And they're not getting that much better yet.

Half a century is a long time, of course, so this is all speculation. But fun ideas can be grounded in reality, and if you learn something thinking about them it's all to the good. -- JoshuaGrosse

The number of hardware connections isn't what's important. AI will never emerge from raw hardware and nobody expects it to. What matters is the number of connections in a software, which can greatly surpass those in hardware. Silicon has a raw speed at least three orders of magnitude higher than meat. This speed can be harnessed by multi-tasking to create a much larger neural net than a straight-forward implementation would allow. In any case, multi-tasking can be done at the hardware level with FPGAs. -- RichardKulisz

Software connectivity is still way below what it needs to be, though, and again I don't think we've been making geometric strides on that front. But you're right, it is a software problem, so gates aren't especially relevant. There isn't nearly as convenient a name for the basic software structural unit, is there?

Afraid not. The basic structural unit depends on how you construct the AI. It can be neurons in a network or frames or whatever.

If technological progress has been following an exponential curve for a very long time, then there will come a point at which technological progress accelerates essentially to infinity. I don't follow the logic in this. The two clauses (either side of the comma) are written as if the former implies the latter, which it does not. Can someone explain please? Gareth Owen

I think the only sense that can be made of it is to take the phrasing essentially to infinity metaphorically.

I and Robin Hansen are two of the somewhat rare Extropians who are skeptical about the Singularity. My personal belief is that while we will create machine intelligence soon, and it will certainly surpass human intelligence at some point and begin creating new intelligent machines, I don't believe this will cause a positive-feedback loop because I don't believe that intelligent beings by themselves invent and create things--only whole economies invent things, often aided by individual beings, who certainly deserve some accolades, but whose contributions are actually quite small in the big picture. --Lee Daniel Crocker

I suppose the question is, looking at it from an economics point of view, how cheap is machine intelligence compared to human intelligence? Right now, human intelligence is far cheaper by almost any measure. But suppose Intel could have the equivalent of 1 Einstein for $1,000. Wouldn't it make sense for them to quickly whip out 1 million of them, for $1 billion? What could 1 million Einstein's do? Even if the individuals among them only contributed incremental gains, the total would be stupendous.

But my contention is that there are only so many incremental improvements that can be made to the current state of technology, and even if we had enough Einsteins to find all of them, they only consititute progress when they build on each other, and that requires communication and organization--and that takes time. I don't argue that the growth will not happen, I argue that a technological economy has an ideal size, much like a business does, beyond which it becomes inefficient, and that the rate of growth will level off before it reaches a singularity.

Am i the only one who thinks we might also be headed for a social singularity rather than a technological one? Or rather that if we have the technological one without the social one it might be a Bad Thing? --JohnAbbe, fond of e.g. Pierre Teilhard de Chardin - The article lists the (technological) singularity as an "information theory" term -- is calling it an information theory term that accurate? I really can't see what it has to do with information theory, in the sense of Shannon, etc. It seems to belong more in futurism (the field, not the art movement). -- SJK

-

The technological singularity is being taken as a given here, and some of you are even expecting it within your lifetime. I'd like to pour some cold water over the whole idea for a moment.

Firstly, let's look at what is true. The amount of "computing power" available for $x doubles every 18 to 24 months. That's been true for almost 50 years, and will almost certainly be true for at least a decade more. Beyond that, it gets a bit hazy, because sooner or later your computers start to get atom-sized limiting how small you can go, and the speed of light limits how fast you can propagate information between the components of a computer.

Now, let's look at what *isn't* demonstratably true. Firstly, the proposition that technological progress has been exponential. That's a difficult case to make - to take a simple example, consider Europe's Dark Ages, or, closer to home, compare the technological and scientific progress of the western world through the periods 1901-1950 and 1951-2000.

But my biggest criticism of this theory is the assumption that Einstein-level Artificial Intelligence is an inevitable consequence of Moore's Law. That's simply not clear yet. Consider this. Computers are a million times faster (or so) and have perhaps one billion times the storage capacity of ENIAC. Has this brought them any closer to, for example, passing the Turing Test? No. Has the extra computing speed available actually helped much? In practice, not really, and in theory, not at all - any Turing-complete computer has got essentially identical abilities, given enough time. Nobody has proposed a scheme for constructing an Einstein that is waiting for a faster computer to make it practical. Looked at another way, current computer chips are probably approximately as complex as the brains of your average insect, but can we replicate the abilities of those brains? I'd argue that we struggle.

Similarly, I spent some time looking at AI research a couple of years ago, and it seemed to me that the area has basically stalled. A lot of obvious ideas have been tried, and whilst having applicability to some particular problems have severe flaws that make them useless for more general application.

The only model we have to work from, the human brain, is still largely a mystery. However, our abilities to investigate its workings have improved a great deal over the past couple of decades, and will continue to improve. Once we have a decent model to work from that fills the gap between our (limited) understanding of how individual neurons work, and the current domains of contemporary psychology (even though both existing domains will probably get a severe revision once we have the completely unknown middle bit worked out), we might get some clue how to construct a real AI mind. When this will occur is extremely hard to say. Research is not a linear curve - progress happens in fits and starts when one breakthrough allows others.

Finally, there is an assumption that, given the technology becomes available, we will make ourselves obsolete with such machines as soon as they are constructed. Given the hullabaloo that gets made about gene technologies, a lot of which replicate gene transfers that occur naturally through bacteria anyway, imagine the ethical fuss that will be made about "playing God" with such machines. Ethicists might well convince politicians and the general public to tie up research for many decades, and restrict such devices to prevent "self-awareness" (whatever that means), restrict their general intelligence, and restrict them as special-purpose devices.

Frankly, even if "The Singularity" happens, I doubt I'll live to see it, and I'm 25. --Robert Merkel

Good comments. I should note, however, that this discussion is based on our binary-linear-logic computers, whereas it is very likely that within 5-10 years they will have been completely replaced with quantum computers (which already exist.) Now, this information is second-hand, so I'm not sure of its reliability, but my friend explained the quantum computer he had read about, and said that it "explored all possible solution paths simultaneously." He didn't quite understand it either, so I'm not claiming that it's a fact, but the possibility must be considered. Quantum computers are very different from regular chip-based ones.
 * I think most researchers in quantum computing would consider your prediction of 5-10 years to be rather optimistic. Very basic quantum computers do exist, but they are not powerful enough to do anything practical. Whether or not the technology can scale to the point where it can actually do something useful is a big open question (the bigger a quantum system becomes, the more fragile it becomes also). Also, quantum computing can give exponential speedups for certain problems (e.g. factorising large numbers), but it is unknown whether they can give anywhere near as big speedups (or are even at all suitable for) general intelligence. -- SJK

Thank you for the clarifications; my knowledge of quantum computing is rather hazy.