I've been following Doug Lenat since he worked with Hofstadter at Stanford in the 70s. Two of the most brilliant people on the planet. Lenat went underground to work on Cycorp in order to insulate himself from lots of extraneous profit-making forces. The fact that he's ready to stick his head above ground is huge.
Consider this. The AI possibilities of Cycorp (yet to be proven) and the data organization access functions of Watson. The combination is the closest we'll get to "skynet" in our time.
Cycorp's goal is to codify general human knowledge and common sense so that computers might make use of it.
It's only a slight stretch to say Cycorp is building a brain out of software, and they're doing it from scratch.
This discussion has been archived.
No new comments can be posted.
Most Ambitious AI Project in the World - 30 years old
|
Log In/Create an Account
| Top
| 11 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(Score: 2) by AnonTechie on Friday July 04 2014, @11:48AM
How many here think that this is feasible or realistic ? I think it may be more of marketing hype than substance !
Albert Einstein - "Only two things are infinite, the universe and human stupidity, and I'm not sure about the former."
(Score: 4, Interesting) by geb on Friday July 04 2014, @11:56AM
They're not trying to claim human level intelligence, or a conscious machine, or anything radical like that. Even the summary only says "closest to" not actually close to skynet.
I don't see anything unrealistic about an algorithm and data set that will try to guess at implied or missing data.
(Score: 3, Insightful) by khakipuce on Friday July 04 2014, @12:14PM
I am very skeptical that isn't anything but and over-hyped knowledge system but I'm sure the market will decide and I'm happy to be proved wrong.
I am confused that on the one hand the article says they are trying to codify all human knowledge and on the other they say that the system has to be taught. So is it a codification of knowledge or an AI that can learn things? Someone is probably going to say "both" but in that case I might as well teach a classroom of people things which they can develop, expand and pass on as spend the rest of my life teaching an AI to do things that humans are already more than capable of.
(Score: 2) by khallow on Friday July 04 2014, @12:22PM
"Both". And there are plenty of teachers. If you have the skills peculiar to "teaching an AI", that can indeed be of greater value than merely being another human teacher. The fundamental problem is that teaching humans is well established and routine. Teaching an AI is at present not even poorly defined. There's such a profound ignorance of what we're doing there, that a few people might as well start making stabs in the dark to see what's not a waste of time.
(Score: 0) by Anonymous Coward on Friday July 04 2014, @12:29PM
Bullshit. Proper teaching to humans is damn difficult, developing motivation and concentration are very hard tasks. People aren't parrots, yet some teachers talk about concepts without profound explaining and not analyzing feedback in an exhaustive way.
If an AI could be converted into the ultimate and tireless teacher, education could evolve and stupid teachers can disappear. Only the best ones would work along with advanced AIs that keep track of everything and assist them.
(Score: 2) by khallow on Friday July 04 2014, @01:33PM
Damn difficult things are frequently made routine and teaching is one such example.
Well, that sounds like a pretty high value reason to at least play a hand or two of the "teach AIs" game.
(Score: 0) by Anonymous Coward on Friday July 04 2014, @12:45PM
If you teach a human, whatever you teach will survive only as long as the human survives, unless that human also teaches it to another human. With an AI, an arbitrary number of copies can be made, all starting with exactly the same knowledge.
I'm not sure whether that is a good or bad thing.
(Score: 1) by fadrian on Friday July 04 2014, @01:10PM
Marketing hype doesn't generally last for more than thirty years (Ted Nelson's Xanadu Project is the only thing that comes immediately to mind).
That is all.
(Score: 2) by carguy on Friday July 04 2014, @01:35PM
Take a look at some of the case studies, links near the bottom of the company website. Looks to me like a very practical approach for certain large applications--and the range of applications seems to be growing.
Rather than lots of hype, they seem to support their company with long term customers.
(Score: 3, Funny) by The Archon V2.0 on Friday July 04 2014, @02:08PM
"Skynet", divorced from what we know about it, sounds like some sort cloud storage gimmick.
"Cycorp" sounds like a factory of blackened concrete where a big cargo door opens and you see a thousand T-800s riding a thousand ED-209s.
(Score: 0) by Anonymous Coward on Friday July 04 2014, @03:27PM
Depends on what you mean by 'teaching'.
Rote memorization of information is mechanical, getting a kid to 'think' is not.
But then, we don't really want people to THINK, do we?
Maybe things AREN'T as they appear, and 'truth' isn't black and white.