5 ms·

Interesting that there seems to be no mention of Cyc, which is explicitly intended to be good at problems like those mentioned.

teddyh 13y ago
But is it any good at those problems?
- seiji 13y ago
  It's best to just let Cyc people continue to think they matter and let them stay busy expanding their knowledge base in the church of first order logic. It'll never go anywhere, but it's better to keep them all contained in one place rather than polluting other fields of study.
  - pnathan 13y ago
    Okay, that was way too nasty and out of line.
  - ScottBurson 13y ago
    See, this is what I mean (see my reply to eschaton). I take it then, seiji, that you also completely disagree with Levesque's program as expressed in section 4.4 of his paper?
    - seiji 13y ago
      Absolutely. 4.4 is a full endorsement of Cyc (and don't get me started on "Cyc" vs "OpenCyc"). Second 4.4 looks a little like "graphplan reborn" or "generic AI approaches from the 80s." Discrete codified knowledge is not the stuff the universe is made of. The problem Cyc will never solve is the "a picture is worth a thousand words" problem. Describing everything in a relational, hierarchal, predicate calculus arrogantly ignores the unrelated multidimensionality of, well, everything.
  - giardini 13y ago
    A merciful end to Cyc funding might be better. Monies saved could be spent elsewhere and termination would free all involved without the expectation that they produce something of value. Although it might cause some grudges ("If we'd only been allowed to add the last 2000 rules, I am certain it would have achieved consciousness!") everybody needs a new start now and then. Had Lenat published working versions or even complete explanations of his previous software (AM, EURISKO) I might be a little more sympathetic. Cyc, the project that never ends, appears to be in a state of perpetual partial implementation, it's most notable product being a constant stream of money into the project. But look on the bright side: perhaps I'm completely wrong, Cyc has actually achieved AI, and the NSA is using it right now!
- ScottBurson 13y ago
  I think we can infer, from Levesque's failure to even mention it, that the answer has to be no. I admit I am curious exactly what Cyc can and can't do, and why. I haven't looked at it closely, though.
  - eschaton 13y ago
    What I've heard is that their system has gotten progressively better over time while their publishing had gotten worse. People in academia can get ResearchCyc and see just what its capabilities are these days; it's basically the same as full Cyc. OpenCyc is a shadow of the real thing, supposedly.
ScottBurson 13y ago
I agree, that is very curious. I clicked through to Levesque's paper, and there is no mention of Cyc there either. I gather Cyc has long been considered a failure by many in the AI community, and perhaps even an embarrassment by some. Still it is hard not to read Levesque's paper as a vigorous endorsement of the goals of the Cyc project, if perhaps not of its exact methods -- see sec. 4.4. I think it is incumbent on Levesque to explain what Cyc has done wrong, if he has a good theory about that -- and if he doesn't, he needs to admit it, because Cyc's ostensible failure makes it seem unlikely that any symbolic, reasoning-oriented approach to AGI could ever succeed.
- rjsw 13y ago
  If Cyc has failed then why do there seem to be so many people trying to build XML schemas that could be used to reimplement it.
  - seiji 13y ago
    Cyc-type approaches are useful for narrow, well-specified domains, just not AGI or actual "intelligence."
    - mark_l_watson 13y ago
      It has been a few years since I went through Ben Goertzel's team's AGI papers, but I find it difficult to believe that Cyc would not be useful for background information for real world knowledge. I would use Cyc, Wikipedia info boxes,etc.

teddyh 13y ago

But is it any good at those problems?

seiji 13y ago
It's best to just let Cyc people continue to think they matter and let them stay busy expanding their knowledge base in the church of first order logic. It'll never go anywhere, but it's better to keep them all contained in one place rather than polluting other fields of study.
- pnathan 13y ago
  Okay, that was way too nasty and out of line.
- ScottBurson 13y ago
  See, this is what I mean (see my reply to eschaton). I take it then, seiji, that you also completely disagree with Levesque's program as expressed in section 4.4 of his paper?
  - seiji 13y ago
    Absolutely. 4.4 is a full endorsement of Cyc (and don't get me started on "Cyc" vs "OpenCyc"). Second 4.4 looks a little like "graphplan reborn" or "generic AI approaches from the 80s." Discrete codified knowledge is not the stuff the universe is made of. The problem Cyc will never solve is the "a picture is worth a thousand words" problem. Describing everything in a relational, hierarchal, predicate calculus arrogantly ignores the unrelated multidimensionality of, well, everything.
- giardini 13y ago
  A merciful end to Cyc funding might be better. Monies saved could be spent elsewhere and termination would free all involved without the expectation that they produce something of value. Although it might cause some grudges ("If we'd only been allowed to add the last 2000 rules, I am certain it would have achieved consciousness!") everybody needs a new start now and then. Had Lenat published working versions or even complete explanations of his previous software (AM, EURISKO) I might be a little more sympathetic. Cyc, the project that never ends, appears to be in a state of perpetual partial implementation, it's most notable product being a constant stream of money into the project. But look on the bright side: perhaps I'm completely wrong, Cyc has actually achieved AI, and the NSA is using it right now!
ScottBurson 13y ago
I think we can infer, from Levesque's failure to even mention it, that the answer has to be no. I admit I am curious exactly what Cyc can and can't do, and why. I haven't looked at it closely, though.
- eschaton 13y ago
  What I've heard is that their system has gotten progressively better over time while their publishing had gotten worse. People in academia can get ResearchCyc and see just what its capabilities are these days; it's basically the same as full Cyc. OpenCyc is a shadow of the real thing, supposedly.

ScottBurson 13y ago

I agree, that is very curious. I clicked through to Levesque's paper, and there is no mention of Cyc there either. I gather Cyc has long been considered a failure by many in the AI community, and perhaps even an embarrassment by some. Still it is hard not to read Levesque's paper as a vigorous endorsement of the goals of the Cyc project, if perhaps not of its exact methods -- see sec. 4.4. I think it is incumbent on Levesque to explain what Cyc has done wrong, if he has a good theory about that -- and if he doesn't, he needs to admit it, because Cyc's ostensible failure makes it seem unlikely that any symbolic, reasoning-oriented approach to AGI could ever succeed.

rjsw 13y ago
If Cyc has failed then why do there seem to be so many people trying to build XML schemas that could be used to reimplement it.
- seiji 13y ago
  Cyc-type approaches are useful for narrow, well-specified domains, just not AGI or actual "intelligence."
  - mark_l_watson 13y ago
    It has been a few years since I went through Ben Goertzel's team's AGI papers, but I find it difficult to believe that Cyc would not be useful for background information for real world knowledge. I would use Cyc, Wikipedia info boxes,etc.