Transhumanism.

I think it will bring both greater capabilities and new ways of thinking, and new ways to interact with each other.
See this essay by Hans Moravec; it is more than a little scary, and probably not the way the future will go; but I expect the future will be as unfamiliar as the one described, but in other, as yet unforeseen ways.
http://www.cni.org/pub/LITA/Think/Moravec.html

 

Log in or Sign up to hide all adverts.

Have you mistaken me for someone else? I am not a neurological expert, theres too much stuff elsewhere I am interested in. I dont really know how Penroses thing is supposed to work. Perhaps you do?

Kurzweil would probably argue that some of them are, what with that Uk researcher whose had a chip implanted in his arm, the work that is done on rat neurons in cultures and little microchips, etc. But my point is simply that Penrose has sugegsted something, but someone needs to test it properly, and has anyone done so?

 

Log in or Sign up to hide all adverts.

The problem is that most proponents of transhumanism on this board are content to wax poetic about their vision of the future without providing a solid reason why it should even be possible. I strongly suspect that conventional computers will be unable to retain or process the information in a human brain, and I also suspect that you may be better off with just the brain.

I only have a very vague idea of how Penrose's theory is supposed to work, but with enough time spent on Google and Wikipedia, I can point you to a document that will explain exactly how it's supposed to work. I want to know if there's an analogous publication for the downloading of minds into computers.

The fundamentals of the Penrose-Hammeroff theory are currently being tested, yes. I don't know if any results have been published yet, though.

Bionic transplants and remote-control rats are conceptually very different propositions from the scanning and downloading of a brain. The former involve knowing how to interface with the brain, and the latter involves knowing to the finest detail possible the brain's inner workings.

 

Log in or Sign up to hide all adverts.

Your right, there isnt as far as I know about downloading of minds. Its premise sounds ok, like Drexlerian nanotech, but geting it to work is another matter. Whilst MRi is still improving, I think we need the next step up from our current data storage technology before we can consider storing the necessary positions. Then as for running a simulation of your brain, based upon your brain structure- thats a lot of work.

As far as I can see, the brain is too complex, to omany feedback loops and outside interferences, to make running it as a program a simple matter. You would have to code for every protein and strand of DNA, then have them interact. your right, it may not ever be possible, but we cannot quite yet dismiss it outright.

 

Baumgarten,

Take a closer look. It has thousands of inputs, a single output, and executes considerable signal processing. Whatever the nomenclature it is a relatively low speed active processing component somewhat more complex than a switch, or in effect a simple microprocessor in the strictest terms.

The essential point was to highlight the parallel processing nature of the brain as opposed to the classical miss-conception that the brain is like a super fast single computer. In this respect the brain is more like the internet, e.g. a very large collection of simpler processing components.

Why not? If the application can take advantage of parallel streams then doubling the number of instructions executed will yield an identical result to two processors running at half the speed. Are you perhaps confusing the inherent non scalability of SMP with the linear scalability of MPP?

 

Um, no. Maybe an ideal SMP machine would yield fully double the performance, but something prevents that - something that also makes the internet unlike a brain - latency.

But anyway, yeah. Brain != computer. My memories are not decoded by programs written in baumgarten assembly.

Please Register or Log in to view the hidden image!

 

The internet analogy demonstrated the independent and parrallel nature of neurons, which was the point, and shouldn't have been taken further. But given a high bandwidth and high speed interconnect between geographically seperate internet devices latency may not be such an issue. The emerging 10Gb and 100Gb ethernet standards would solve much of that, along with RDMA. But the connections between neurons, the synapses, do have significant latency since those connections are not electrical but bio-chemical and are relatively very slow.

Pure artifical electrical neural networks should be able to outperform bio-brain equivalents quite easily. But the brain is not SMP, which is also my point, it is MPP, which means building hardware to create an artifical brain is entirely within our reach with only modest advances in current chip speeds. I.e. we do not have to break the laws of physics to achieve the power needed, or invent exotic new futuristic processors before human equivalent AI becomes viable.

 

baumgarten,

Ah I see you do not know much about parrallel processing. Processors in an SMP have shared memory which becomes a point of contention. That bottleneck increases exponentially as the number of participating processors increase. An MPP has no shared memory and can achieve linear scalability, much like the neurons in the brain.

 

You're right, I don't know much about parallel processing. I'd never heard about MPP before reading of it in this thread.

From what I understand, neural networks are very capable today for their particular purposes, but they could not simulate a brain, since there is much more diversity in the nodes of a brain than in a neural network. Not all neurons follow the "many inputs, one output" model, for example; and then there is also the question (still unanswered, I think) of whether quantum mechanics plays an important role in the functioning of the brain.

Personally, I think this is the wrong route to go down for brain enhancement. To me, it seems that the field of genetic engineering offers much more promising prospects than computer science, if for no other reason than the assembly of the brain is taken care of by nature, leaving us with only the design. I think that even the concept of merging with machines proposes reinventing the wheel in a way.

 

Baumgarten,

You may be right that enhanced and carefully directed genetic engineering could result in enhanced human intelligence. But this seems likely to be a very slow process, taking hundreds of years and multiple generations. Well, perhaps that is OK, after all, what's the rush?

However, computer technology has been doubling in power every 18 months consistently since the 1940's and shows no real sign of ending in the near term. And machines are beginning to increasingly appear intelligent, limited primarily by raw compute power and more importantly by appropriate software algorithms; aspects that will certainly rapidly change in the near term since they are also closely related.

The expectation is that computer technology will achieve greater than human level intelligence long before we can achieve the same thing through genetic engineering. It is also expected that this intelligence will feedback into further AI enhancements, i.e. the rate of increase in machine intelligence will further rapidly outpace any chance for us ever keeping up through biological means.

If this plays out as expected, and as I feel is inevitable, then the arrival of such super intelligence means we will cease to be the dominant species on the planet. We either accept that or learn how to use the same technology to enhance ourselves so we can join the AIs on an equal footing with equivalent advancements.

As for the quantum claims: That’s a pure red herring. Intelligence operates at a macro level; even low intelligence lifeforms use the same neural mechanisms with the same quantum features which clearly do not make them any more intelligent. The difference between us and them is the complexity of the neural networks which at an appropriate level generates self-awareness. Whether the quantum effects are required for bio building blocks is irrelevant, providing we can generate the appropriate processing complexity through other means, e.g. silicon, or whatever, then the quantum / bio dependencies become a complete non-issue.

Now will the computer based brain equivalent processing engine have identical characteristics to the bio version to allow absolutely identical copies during mind uploading? I doubt it. I’d hope the differences are sufficiently minor such I’d be willing to trade them for the open-ended lifespans that a non-bio lifeform would provide. In the end I’m only interested in survival and biology right now doesn’t offer that and doesn’t look as though it will be able to compete with AI’s in the future.

The outlook I see is that we join them and upgrade or we become extinct or entirely sidelined.

 

Baumgarten,

I'd prefer the analogy that nature has created a cartwheel and now we have the opportunity to turn that into a formula-1 racing wheel. It is not a matter of re-inventing but more about moving off the first rung of the evolutionary ladder and climbing much higher than nature alone could achieve.

 

I take issue with this position, and I'll cite Gordon Moore himself (a quote found in the Wikipedia article on Moore's Law) to explain why:

The population keeps climbing as well as microchip R&D budgets. Even if technical obstacles are overcome, there will be economic obstacles that will not be so easily surmounted. Eventually it will become more important to get food and shelter than to buy a computer, and at that point funds and motivation for developing faster hardware will become scarce. Technology is not the driving factor behind technology.

There is such a thing as "fast enough" - personal computers haven't changed much in the last year or so. There's currently no sufficient economic incentive for 4+ GHz personal computers, or massively parallel home machines.

What is "human-level" intelligence? What are the different levels of intelligence? How do we measure it? What do you mean by intelligence, anyway; just the ability to calculate?

There are certain aspects of quantum effects that are thought by some to allow us definite advantages over artificial intelligence, such as the ability to reach valid conclusions through non-computable means. It is far from a red herring.

What makes you think that a biological lifeform is incapable of providing an open-ended lifespan? There are tortoise species with just that.

As for being sidelined, I completely disagree. In evolution, species become more relevant with age. Even if it can think faster than the average human, a species of intelligent machines would have a lot to prove. And we're the ones with the weapons and our fingers on the ON/OFF switch.

 

Baumgarten,

What was your point of the Moore quote? I did say “near term”. You and he appear to be agreeing with me.

PC’s are not driving computer maker plans. PC’s are a relatively new phenomenon, in essence they have come and gone and the major computer makers are returning to normal work, i.e. the business enterprise which is demanding more and more computing power. The PC is rapidly becoming just a connection to the massively expanding internet where the real compute power is held. But the BI (Business Intelligence) market is huge and only now are we seeing businesses pick up the massive systems needed. While your perspective might be the PC the real driving forces are elsewhere.

I’d imagine you can tell the difference between human intelligence and a dog’s intelligence, right, without too much trouble? Feel free to develop some math to quantify it, but the gross differences should seem very clear to most people.

And those working in AI disagree. History is littered with perceived experts who claimed key achievements were impossible only to be shown completely wrong by the technologists. I don’t see any practical obstacle yet that prevents us from developing AI software and hardware. At this point it appears to be purely an engineering task, with inevitable design issues to be resolved.

I don’t. I’m a strong supporter of anti-aging research and I will use cryogenic services if I’m unlucky enough to die. I simply see that a biological future is far too restrictive and fragile and that there will be far better alternatives.

They do? Don’t think you have a clear idea what you meant by that.

For example? And by faster I mean 10 times, 100 times, and more. I don’t think you quite comprehend what we mean by super-intelligence.

LOL. Once the algorithms are out, that’s it; you will not be able to put them back. There will not be a single target. And worse, you will have to fight the many people like me who want and are making it happen.

 

There is no reason to expect that future artificial intelligence will not operate on the quantum level as well; if quantum-level effects are what it takes to allow self-awareness then quantum-level effects will be used.

The idea of enhancing intelligence with genetic engineering is an intriguing one; I personally think that such genetic manipulation with very specific goals will need very powerful computation to predict the results of such manipulation beforehand. Very, very powerful machines dedicated to predicting gene expression will be necessary, although these numbercrunchers would probably not need to be self-aware.

But assuming genetic manipulation of intelligence were acheived (which I think it will) then the result will probably be humans with larger heads. It may well be possble to make humans smarter without increasuing their brain capacity- up to Einstein level or beyond; but to continue to increase human intelligence using the biological route will eventually require larger brains.

Eventually the brain will need to be carried about in a truck, or housed in a building. Anyone read Last and First Men by Stapledon? The Fourth Men were encased in brick-built skulls and tended by human sized slaves. Perhaps that will come...

But some of the visionaries considering AI imagine much larger constructs. Anders Sandberg here describes brains the size of Jupiter - pdf
while Robert Bradbury here describes solar system brains.
http://www.futurehi.net/docs/Matrioshka_Brains.html

I can just about imagine artificial structures on this scale, but it is difficult to imagine biological brains this big.

 

eb,

Yes interesting perspective.

 

Mu tuppence worth, on a tangentiallyr elated subject- I dont think that the evolution of processing power right now is being driven by PC's, but by network portable processing functions, like your mobile phone, (Which probably has as much power as the space shuttle or suchlike) or your car. What mobile phones require is small light processors with low power needs, which means that they are still driving forwards the technology necessary to make more powerful computers. Then in a few years time virtual reality and interactive stuff will become more of an issue, and your home computer will need to be more powerful again.

 

I agree. That's why I say that technology is not driving technological development. It's an economic impetus.

 

Did you read what I wrote under it?

I cited personal computers only as an example to illustrate the way computer manufacturers behave in general.

That doesn't really answer any of the questions I asked.

I never did call it an impossibility, but I did bring up some research that suggests that algorithmic computing might not get the job done.

How come it's too restrictive? What's holding it back?

Maybe I didn't make it clear enough to you. A species demonstrates its viability through longevity. Every species so far has been carbon-based; as I said before, a species that is brand new, not to mention not even carbon-based, however intelligent, has a lot to prove with regards to its viability as a species.

I'm not sure that you completely comprehend what you mean by super-intelligence, either. It seems like a blanket term to express a vague, underdeveloped idea to me.

Which are they? Tools or competition?

 

This is silly. You misunderstand how we determine intelligence when studying animals. It has nothing to do with processing power, as you probably think, but rather behavior. In other words, intelligence is not a substance or quality that one individual has more of than another. Intelligence is about behavior, not computing power. It's seems to me that people talking about artificial intelligence think that intelligence is determined by how many calculations per second the brain can perform. Looking at it in that way, it's easy to think that one day computers will be more "intelligent" than human beings, because they will be better at information processing. They already are faster, they just lack the parallelism, but that will come soon. The point is, we look at the behavior of a dog and say that the dog is less intelligent than a human being because we are putting the dog into the context of the human niche, i.e. Homo Sapiens' self-imposed environment. If we put the dog into another setting, say, the Serengeti, then that animals behavior seems to be more intelligent than your average nuclear physicists'. Consider it for a moment in this light.

 

Jaster,

Your assertion is significant gibberish in virtually every aspect.

Intelligence in the context of this forum is essentially the ability to think and reason and to reach and exceed the status of "I think therefore I am", i.e. the realization of self-awareness.

The other primary factor that distinguishes lower forms of life in terms intelligence is the correlation with brain complexity in terms of both size and the number of folds, and that translates essentially to processing power.

Also your reference to a computer being faster is simply ambiguous.