How long to human equivalent AI

Kmguru has stated that he can’t see uploading taking place for perhaps another 150 years. I think it will occur much much sooner than that, within then next 20 possibly but certainly within the next 50.

The first step is to create the machinery that has at least the same processing power of the human brain. When we can do that and push that technology beyond human intelligence then surely the problems of uploading will be resolved at a rapidly increasing non linear rate.

Human processing power equivalence by 2012.

Moore’s law says that computing power approximately doubles every 18 months. This has held up almost perfectly since 1940.

A human neuron fires at around 200 times a second (200Hz), and there are 100 billion of them in the human brain operating in parallel. A powerful PC at present can operate at 2GHz, or the equivalent of 10 million neurons. That means we would have to link 10,000 PCs together to achieve human brain equivalence. Well we can’t quite do that yet. But in 10 years we should have a computing chip of around 200GHz, and at that point we only have to couple 100 of them together to achieve human brainpower. And that we can easily achieve, i.e. by 2012 we will have computing power to rival the human brain.

The Software.

But computing power alone will not be enough; we also need the software. And that should also not be a problem. As CPUs have become increasingly powerful recently we have seen a corresponding increase in the interest in AI software. The best approach, I still believe, is to emulate the neural networks of the human brain, and that is actively being pursued.

A few years ago it was difficult to find any books on AI in regular bookstores. Several weeks ago I checked again (9/2002), and not only are there more books there was a whole rack dedicated to AI.

It seems inevitable that human equivalent machine intelligence will be with us within the next 10 to 20 years. Assuming their intelligence is based on human neural networks then it seems very likely that they will be self-aware and our equals.

And Beyond.

But technology will not be standing still. If Moore’s law continues to hold then within a further two years their intelligence will be double that of humans. And at that point we have no way to imagine what will happen next. The emergence of super-intelligence means humans will cease to be the dominant intelligence on the planet.

 

Log in or Sign up to hide all adverts.

Cris

The brain processes different types of information simultaneously from the various senses; images from the eyes, sounds from the ears, etc. How would you incorporate these functions into an AI platform and would you expect this advent of technologies adding to your time-lines ?

I would think that the processing of images alone would be a massive undertaking of technological engineering.

 

Log in or Sign up to hide all adverts.

Log in or Sign up to hide all adverts.

Have the events of 9/11 etc etc slowed or even accelerated Moore's law? Kmguru is right, to the extent that global events and ideas influence technological development (if we were still in the space race we might be to mars by now). Likewise, when the "religious people" start speaking out about the evils of mind uploading and a.i etc the process will likely be slowed, directly or indirectly.

But I wouldn't dilute the time of these things being as commonplace as game boys to 150+ years, I'd give it seventy, tops. But what do I know?

Please Register or Log in to view the hidden image!

 

Hi Q,

Image processing is a significant issue. Image capture is not the issue; that we can do quite well, it is the interpretation of the image that is the problem.

The human brain holds a large knowledge base that allows it to interpret what it sees. Some image processing machines can capture images and can make some sense of what it sees but to be truly effective it must have that all important knowledge base plus a massively powerful processing mechanism that can search the database with sub-second response times.

This isn’t so much a question of complexity but of raw power. Note also that the human brain does not hold detailed images but only vague outlines. The brain is then very good at filling in missing details.

No, the time-line doesn’t change, the key factor is that all important increase in power that we should achieve if Moore’s law holds true.

 

Using Moores Law and stretching convential techniques we should have the hardware in place in 20 years and i guestimate another 10 years of "DNA like breeding" of software (nature has programmed in millions of years) before SKYNET will wake up and take over the earth....

But maybe the latest developments in quantumcomputers can speed up that process. We are already at 7 qubits capacity and the use of josephson switches promises the use of thousands of interlinked entangled quantum dots, each capable of millions of quantum states...


It's only Sci-Fi until some dude build it in his garage...

 

I think the scenario Cris outlined is possible but is highly unlikely. Why? Look at the debate that has slowed down stem cell research. That is peanuts, actually much less than peanuts, when compared to some of the public backlash we will get from developments in AI. There will be so many strangeholds and restrictions put on R&D...if we can develop human level AI in less than 50 years I will be very (pleasantly) surprised. I may be wrong, but I think once people start seeing just how close we are to reaching some incredible milestones, they are gonna freak out.

 

fading,

I think you are only viewing this issue from the perspective of the USA. In the UK for example there are virtually no limits on stem cell research and the government there are actively encouraging such research. The same is true for many other countries. The issue with the USA right now is that its government is dominated by the religious right. A democrat led administration might change that but not by much. The USA will simply lose the lead in anything that results from stem cell activities.

Fortunately, for AI, the computing power in hardware will arrive first, and likely before anyone quite realizes what is happening. Once the technology is out then it will be impossible to put back in the bottle. But it will be the software that will be key to AI, and anyone anywhere in the world can write software. China and Asia are rapidly recruiting and training programmers in droves and we in the USA are actively given them a lot of our work because they work for much lower salaries.

The other fortunate thing about AI taking on human level intelligence is that religionists will probably believe their own propaganda and refuse to believe that man can create anything superior to man; and man is God’s ultimate creation, right?

And by that time, hopefully, the software, or its seeds will have been copied a trillion times all over the world. After that the USA will be out of the picture and mankind had better wake up very fast.

At least that is my optimistic view.

 

Yep you are right. I wasn't aware that such stem cell research was occuring in UK. Good to know.

I used to be one of those programmers hehe. Don't forget India...lots of talented programmers there doing what I used to be. Could be a serious problem for US in the future.

Maybe so...it would be interesting at least...

Please Register or Log in to view the hidden image!

Are there any good books out there on this topic? Can be sci-fi but I would prefer fact-based...

 

Wars do tend to cause a rapid increase in technology.

 

What about storage media? How much information, in the form of memory, does the human brain have? RAM=short term, ROM= long term?

AI would have massive implications in the manufacturing world, and this would tend to override any qualms by anyone else, as money talks. However, a thinking machine would effectivly replace a human worker, cheaply.

Im not sure, based on my very limited knowledge of AI, that a machine could think. Remember? yes, they do now. Problemsolve? Maybe. Help me out here. The nuances of human thought are gossamer. AI would not be intelligent, as in human intelligence with out emotion to guide that intelligence, just a cold computer.

Maybe I need to read some more. Got any links?

 

Two points:

AI research and medical research are two different animals i the eyes of gov't regulation, aren't they? While one gets slowed down by red tape, I see AI being developed and pushed through BY the gov't. It's good to know that the UK is continuing in stem cell research.

And, while I'm fully for a positive development in AI and then true machine intelligence, there IS a danger, is there not? Joking aside about SKYNET/Matrix/etc, they aren't an impossible scenario. Not likely either...usually the future throws a curve no one saw, but just as nanotech has their gray goo potential, AI has the problems with potentially overrunning its creators before they know it.

Pessimistic or realistic?

 

The problem with having "human AI" is not really in the hardware. It's the software. We can't even get computers to have equivalent vision sytems as people.

I think we may have very powerful computers soon, but no so human like. as was said, processing power is not enough.

 

I can build them now. But it is not AI. AI never even started to work.

 

robotic soildeir {militant}

In an effort to acquire and maintain (or better yet, transcend the realm) of the edge that military powers (as the United States) possess, by not only being the pioneers and leaders in, the very field of, innovative military strategies that (in turn) gaved them opportunity to have that unfair advantage, but also by staying
two-steps ahead of any potential counter-reaction to their war strategy innovations, our company have this unprecedented ingenuity, which comes now, as an opportunity for these very powers to cater to in reasuring that very 'short-of-ultimate' global power "two steps ahead" edge.

 

did you catch the snippet in Sciam about the guys with the slices of the brain and the 20 years worth of nueron level mapping and the bizness?

 

Chris,

<i>"brain does not hold detailed images but only vague outlines."</i>

Justify this statement please.
as i have heard Sub-consious brain holds every detailed form of data including imaging,only fact is that recalling power of that memory chunk is poor,isnt it?
just curious...



bye!

 

ZION

sir,your outline is,your image?in other words,the brain do hold detailed imagery!!

Please Register or Log in to view the hidden image!