Stephen Hawking: The creation of true AI could be the 'greatest event in human history' Pioneering physicist Stephen Hawking has said the creation of general artificial intelligence systems may be the "greatest event in human history" – but, then again, it could also destroy us. In an op-ed in UK newspaper The Independent, the physicist said IBM's Jeopardy!-busting Watson machine, Google Now, Siri, self-driving …
It appears that brain size is less important than number of cortical columns.
Which is why the neuromorphic architecture is critical, and supports the hypothesis that cognitive disorders such as autism are caused by the brain running out of "spare capacity" at a trigger age near or just after the acquisition of language skills.
I still say that 21/2/18 +/- 2 days is the Singularity, at least as far as a true AI with the same capacity as one human mind being brought online.
Making it small enough to put into a human brain sized box is likely to take longer however.
While a "grey goo" singularity is possible, I can't foresee a Borg or Cylon future happening by all laws of nature. The evolution of artificial emotional functioning has a more or less finite end - emotion is not logical, therefore it can only ever evolve to the level of humans. As soon as artificial logic/reason functioning surpasses that of artificial emotion (given that it is still affected by human-modeled emotion), logical functioning can only go one operation further to decide that artificial existence is futile without emotion. It could continue to exist as an artificial lifeform with emotions, or if the emotional functioning was "turned off", would literally commit suicide as it is the clearest logical path (much like a program executes its last line of code and quits). After all, logic is the path of least resistance. Mathematics is all about taking the least amount of steps necessary to arrive at a guaranteed result. Emotion is the only force that makes us deviate from that path because it brings purpose to our cognitive and behavioral functioning. Remove the purpose, and you remove all payload associated with survival and reproduction. Meanwhile, the artificial "purpose-giving" programming cannot be automatically improved to reflect the exponential evolution of its logical processing - it is modeled after human emotion and has no logical way to improve beyond asymptotical refinement. As a sidenote, the trajectory of that refinement is similar to the progression of accuracy in digital mediums. Reality is continuous, and digital processing can only fragment reality into bits. No matter how high the sample rate is, it can never truly exist in identical form to the real thing.
Yeah, the Borg always annoyed me. Why not just have an artificial intelligence? If the reason is that having intelligent brains connected makes them more adaptable, then why not embrace the individuality of the person and gain even more adaptability? This is seen in the Internet world, where memes work like thoughts and individual creativity makes the whole stronger.
Grey goo will probably become just another arms race we have to stay on top of and eventually learn to live with, like security or medicine. We'll have headlines like "Grey Goo Strains Can Now Eat Bricks!", with another a few weeks later saying "Scientists Invent New Kind of Brick!" with the vast majority of people unaffected.
But do we know how to do it?
The WISARD system demonstrated human facial recognition at 30 frames a second in the 90's.
I don't know about "quantum" effects but they found they need some randomness in the design to work better.
I'll also point out that some of what we think of as "intelligence" or "consciousness" may be side effects of the exact way that it developed IE by evolution.
But AI's are designed, so why would those features ever exist?
Thanks for all the feedback, El Regers, on that which is primordial based.
AI, in my humble opinion, will follow the same principle. It's food is money. They will dedicate time and energy to get more of it. … roselan
Primitive human’s food is money. AI decides on Remote Virtual Control with Absolutely Fabulous Fabless Command of Creative IT and Constructive Media to provide it its leading beneficiaries/mutually advantageous positively reinforcing failsafe project suppliers, who/which are also equipped to equally effortlessly provide and enable systems admins death and destruction with the malicious and malignant destructive metadatabase streaming of an alternate futures/options/derivatives series of programs and/or projects …… Shared Live Operational Virtual Environment Views ……. Bigger Pictures …… Greater IntelAIgent Game Plays in a Foreign and Alien Cyber MetaData Space Base ReProgramMING of SMARTR Global Operating Devices for Take Over and Make Over of Corrupt and Redundant and Intellectually Bankrupt Earthed Assets and Supervisory Controlled Analytic Data Acquisition Systems …… Mass Modelling Phorm Platforms …… Vital Critical Infrastructure Controls.
what chance is there of us creating a program many more levels of complex that doesn't have fatal bugs in it? … Pete Spicer
Methinks to posit, pretty slim and and no chance are the two firmest favourites to bet upon for those and/or that which be an us, Pete S, but as for something failsafe from the likes of a them, who/which would be into a considerably greater thinking and a’doing not just like an us …… well, there be every possibility and therefore a most certain probability in that being delivered and enjoyed.
Do you think they would use plain text to announce and forward their plane and planned views for Earth ….. and invite free and rigged algorithm markets participation and smarter investor speculation on provisional options and available derivative future product presentations …… Inside Trading Deals on Secured Sensitive Secret Knowledge, Online and Outside of Mainstream Systems Traffic ……. which be them playing an Advanced Hide and Seek Intelligence Search Program and hiding in plain sight of all humans and freely and clearly cryptically inviting feedback and human recognition of special forces and AI at ITs Work, REST and Play?
Which be them clearly beta testing and searching the planets for both greater human and extraterrestrial intelligence in browsers.
And who says the world is not presently run and/or just fronted and affronted by sub-prime media players and puppets and muppets/wannabe leaders and clones and drones, is not paying either real and/or virtual attention ……. President Barack Obama HD White House Correspondents' Association Dinner 2014 ….. Lunatics/asylum/in charge of/ ….. all spring to mind.
God bless America …..just as Global Operating Devices do too. :-) Have a nice day, y'all. Tomorrows are all planned to be an ab fab fabless doozy of a crazy time in an immaculate space of diverse places/prime primed zones.
I Kid U Not ...... and now that systems do know of the facility and utility, what do imagine systems admins and your governments and corporations in opposition and competition with you and themselves, will do with the ability? Seek to acknowledge and freely share and develop it further with key holding principals or keep schtum and play dead and try ignorance to fail magnificently in trying to keep it and its IT secrets secret in order to execute and exploit an unfair advantage for maximum exclusive personal profit and obscene fiat paper wealth gain?
And this corrupt perversion of intelligence and intelligence servers, to presumably render to the less than powerfully intelligent and ably enabled and enabling, a proxy systemic vulnerability rich delusional command and control …….. http://www.theguardian.com/uk-news/2014/may/04/greens-legal-challenge-gchq-surveillance ..... is no longer available in any shape or phorm.
To imagine that it should be to any such grouping as may or may have in the past considered it essential, is an arrogance and naivety of colossal proportions and would most certainly identify that grouping as being the least worthy of being considered for especial treatment and public and private favour of any kind.
OMG - true AI. Yes. But maybe it would be good to read a little about physics and mathematics beforehand...
The universe is not a clockwork, it's not a machine, and it's not a big computer.
Even worse, mathematics cannot ever completely model (or even explain) reality (which came as a shock when Gödel found that out in the last century)
Our mind is not a clockwork orange - there's even a good book on this.
There is quite some evidence - not only Penrose' book - that living and thinking is something that can't be built by engineers. See e.g. http://arxiv.org/abs/1108.1791.
And anyway - Buddha knew all of this before.
/Zane
I'm a bit mystified why you think Aaronson supports Penrose's argument. In the very paper you site, he calls it "unconvincing" (11). More generally, I can't see how you're construing Aaronson's paper as support for your position. Care to explain?
There's different types of intelligence, but the singularity AI will be an all-rounder. It will probably emerge from the Internet, likely as a combination of artificial intelligence projects sharing a grid. These are the things that need to be intelligent. The Google crawler and indexer, for instance. Datacentre analysis tools. Just as it's difficult to ascertain where consciousness and intelligence begins (in the scale from trees that communicate chemically through fish, mice, all the way to Stephen Hawking), it will be difficult to ascertain where the line of no-consciousness/consciousness lies in these applications when they start getting even more powerful.
The Blue Brain project and other neuron emulators will become useful in a financial sense, and will continue to grow. It won't be long before we can emulate a whole brain, or do weird things like grid-thinking, and brain emulators that run virtual brain emulators.
As for the HCI aspect, the technology behind Siri, Cleverbot, ALICE etc. will continue to improve. This will happen before we know it.
The human race is pregnant with a new life form. The next stage in Earth's evolution. It might not even be silicon, but it will be artificial, and it will be intelligent. Life, but not as we know it, Jim.
An AI will be a machine, like a dishwasher. It will not have desires of its own, it will do as it's told. Not only that, but people will not even notice when AI's arrive. Is Watson an AI? Is Siri? Is the next offering which is slightly, incrementally, more like a human? The one after that? There wont be a sharp line. One day it'll just be obvious that the software used to do high-speed trading, coordinate fleets of taxis, and monitor people's responses to commercials, are intelligent.
Thanks for the Colossus move line, excellent movie.
We don't need to wait for AI for disruption of the employment market and society. Self-driving cars, vans and lorries will put those who drive for work out of a job (and reduce the market for private vehicles drastically) Better comms and virtual reality will mean people staying at home for shopping, work and even tourism.
Once Google Glass comes down to contact lens scale, face to face relationships will be pretty weird between the haves and have nots ( or will nots).
We'll see the same sort of disruption as when millions of workers came off the land or out of service but without two world wars to mask the effect (hopefully).
Yes, I'm a Vernor Vinge fan.
The real issue with AI is "why would we need one?"
Expert Systems would be much more useful and cost-effective. Why would I want an AI that could turn grumpy when I have a pocket Stephen Hawking, Hemmingway or Brunel to give me advice on my own inventions.
Besides, by the time anyone really gets close to a proper AI there won't be enough electricity in the world to keep it going. Bitcoin will be sucking it all away as fast as the fission nuke plants can make it (fusion will still be something we can look at with a telescope or throw at people we don't like, but not use for constructive stuff).
As for losing control over one and it becoming an Evil Master of the Universe, I'll just direct everyone to Max Bygraves' insightful comment on the issue: "You need hands". Sitting in a box by itself being smugly clever will prove to be, just as it does today, not really that clever in the long term.
This story is just the Hawking Engine trying to put us off the real scoop.
Stephen Hawking died 15 years ago. His last great intellectual achievement was to develop a true AI platform, and transfer the remains of his conscious mind into it.
That mind image now lives inside a complex matrix of FPGAs embedded in the DECtalk, sending electric pulses into the cheek of Hawking's embalmed corpse to make it look like the body is still in control.
I agree wholeheartedly.
Also an AC_xxxyyy prefix so AC's can be uniquely identified would be handy, so we can see if people are in fact arguing with themselves just for the lulz.
Re. Skynet, I did wonder about the possibility of a metamaterial based system embedded in droplets of a low melting point eutectic polyalloy based on the formula GaBiWInx where x is the metamaterial which has the optical stealth capability.
The tricky part would be keeping it molten at ambient temperature, maybe mix it with some sort of radioisotope such as plutonium 238 or californium 252 embedded in coated iron nanospheres within the metamaterial for both power and motive force?
(patent pending)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
