BBC science reporter, Boston
Machines will achieve human-level artificial intelligence by 2029, a leading US inventor has predicted.
Humanity is on the brink of advances that will see tiny robots implanted in people's brains to make them more intelligent said engineer Ray Kurzweil.
He said machines and humans would eventually merge through devices implanted in the body to boost intelligence and health.
"It's really part of our civilisation," Mr Kurzweil said.
"But that's not going to be an alien invasion of intelligent machines to displace us."
Machines were already doing hundreds of things humans used to do, at human levels of intelligence or better, in many different areas, he said.
Man versus machine
"I've made the case that we will have both the hardware and the software to achieve human level artificial intelligence with the broad suppleness of human intelligence including our emotional intelligence by 2029," he said.
| | We'll have intelligent nanobots go into our brains... to make us smarter Ray Kurzweil |
"We're already a human machine civilisation, we use our technology to expand our physical and mental horizons and this will be a further extension of that."
Humans and machines would eventually merge, by means of devices embedded in people's bodies to keep them healthy and improve their intelligence, predicted Mr Kurzweil.
"We'll have intelligent nanobots go into our brains through the capillaries and interact directly with our biological neurons," he told BBC News.
| | CHALLENGES FACING HUMANITY Make solar energy affordable Provide energy from fusion Develop carbon sequestration Manage the nitrogen cycle Provide access to clean water Reverse engineer the brain Prevent nuclear terror Secure cyberspace Enhance virtual reality Improve urban infrastructure Advance health informatics Engineer better medicines Advance personalised learning Explore natural frontiers |
The nanobots, he said, would "make us smarter, remember things better and automatically go into full emergent virtual reality environments through the nervous system".
Mr Kurzweil is one of 18 influential thinkers chosen to identify the great technological challenges facing humanity in the 21st century by the US National Academy of Engineering.
The experts include Google founder Larry Page and genome pioneer Dr Craig Venter.
The 14 challenges were announced at the annual meeting of the American Association for the Advancement of Science in Boston, which concludes on Monday.
http://news.bbc.co.uk/go/pr/fr/-/2/hi/americas/7248875.stm
4 comments:
A pretty bold claim... and that's coming from someone who's primary field of study was artificial intelligence.
Now, artifical intelligence that with human-like intelligence? Well, that's a bit of a loaded statement... That depends entirely on the task we're giving it to accomplish. A thermostat can make "decisions" based on sensory input it takes in from its environment, a task a human could perform quite handily. Does it make intelligence decisions? As long as everything is in working order, yes. In fact, it's much better at recognizing what the ambient temperature is than a human can. And that's great, but that hardly qualifies it as a human-level intelligent agent.
Simply getting an AI agent to recognize the phonetic difference between "sh" and "zh" is retardedly difficult.
Human beings are good at absorbing massive amounts of sensory data, and spontaneously picking out the important parts. When you walk over to the fridge and pull out a beer, a human brain doesn't ask itself "is that a tree? no. is that a sofa? no. is that a cat? no. is that a black rubber dong? no..." Our "neural network" is able to extrapolate almost instantly from past experience that this is a can containing beer, that it's likely to be cold and delicious, and that you can consume it mightily.
Machine intelligence, on the other hand, is absolutely awful at this kind of stuff. The very nature of the beast means that we have to do exactly that kind of comparative iteration at some level to recognize what it is we're looking at, even with an artifical neural network.
In my opinion, human level intelligence, along with its diversity and adaptability, is an emergent property that falls out of the mechanisms we have to collect and make sense of sensory input. But that's a subject for another discussion... In any case, for the purposes of this article, there are certain obstacles to our advancement in certain technologies that, until overcome, won't be able to provide the fidelity of sensory data that would make this level of artificial intelligence possible. And I think 2029 is a bit optimistic.... in the same way that Flying Cars in the year 2000 was optimistic.
Example: The accuity of the humjan eye is amazing, with a resolution of data points FAR higher than anything we can simulate with a camera. Now, to make sense of input from a camera, you have to iterate through each sample point (think of it almost as a pixel on a monitor, for instance) to determine what colors have been sampled before we can make sense of the picture. And we have to do this somewhere around 30 times a second to match human visual accuity.
Well, this is a problem. Every time you increase the resolution of the sampled image, the processing power required to be able to iterate through the sampled points in a reasonable amount of time sky rockets. If a camera samples an image at 480x320, that's 153600 sample points... (153600 inputs to a neural network, in some models)
If we crank that up to 1024x768, our sample set suddenly becomes 786432. And 1280x1024? 1310720.
Can raw memory capacity and processing power overcome this? Well, it hasn't yet, despite Moore's law.
But whatev. In the end, all I'm saying is, take this with a grain of salt. When I have an autopilot for my car that can take me wherever I want to go, I'll be impressed... but I'll be convinced that it has matched mankind in brain power when it can tell me what its favorite color is.
Yeah, Kurweil has been saying this since before 2000. He really lays out his theory on it a more clearly in the book The Age of Spiritual Machines. It has a lot to do with the Law of Accelerating Returns and R.K.'s boundless optimism :-)
You should get in touch with Matt (you the other contributor to this blog Matt). He is doing grad work in the hardware aspect of AI I think.
Yeah, I remember his cockroach stories! Fascinating stuff! I'd love to be involved with robotics in some fashion... UNCC does have a robotics program, but now I'm too busy making AI brains intended to entertain us, not do, like, useful stuff. :P
Watch out Matrix - here we come.
Post a Comment