Wrong Question asked out of ignorance
on
Where's HAL 9000?
·
· Score: 5, Interesting
These sorts of articles that pop up from time to time on slashdot are so frustrating to those of us who actually work in the field. We take an article written by someone who doesn't actually understand the field, about an contest that has always been no better than a publicity stunt*, which triggers a whole bunch of speculation by people who read Godel, Escher, Bach and think they understand what's going on.
The answer is simple. AI researchers haven't forgotten the end goal, and it's not some cynical ploy to advance an academic career. We stopped asking the big-AI question because we realized it was an inappropriate time to ask it. By analogy: These days physicists spend a lot of time thinking about the big central unify everything theory, and that's great. In 1700, that would have been the wrong question to ask- there were too many phenomenons that we didn't understand yet (energy, EM, etc). We realized 20 years ago that we were chasing ephemera and not making real progress, and redeployed our resources in ways to understand what the problem really was. It's too bad this doesn't fit our SciFi timetable, all we can do is apologize. And PLEASE do not mention any of that "singularity" BS.
I know, I know, -1 flamebait. Go ahead.
*Note I didn't say it was a publicity stunt, just that it was no better than one. Stuart Shieber at Harvard wrote an excellent dismantling of the idea 20 years ago.
[snip]... "junk AI". These are the branches of AI that exist more because the metaphor is compelling rather that the results or prospects. These include: Neural Networks, Genetic Algorithms,... [snip]
I suppose that your own articles written on 2nd order neural nets were part of your 'junky period' then, right?
Yep. And I paid for that too. While I was doing my PhD, someone at MIT was doing very similar work, but instead of using 2nd order NNs because they were cool, he had formulated his work with a solid mathematical base.
Guess whose dissertation was better received.
In my defense, I started in a time when the whole world was gaga over NNs and I was swept up in the hype. That's why I (like the ancient mariner) roam the earth issuing warnings to others.
Well, lots of reasons, the simplest being that it's a hard problem. But that's a cop out.
One issue we've had is that because intelligence is an observed phenomenon, not a defined one, its easy to think you're much closer to a solution than you are. The usual process is to observe intelligent behavior, and try to infer a formal problem from that to then try to solve. That problem eventually gets solved, and we discover we didn't ask the right question. Each failure has moved us closer in many important ways, just not directly at the target. It's a like a predator unsure of exactly where the prey is, circling and closing in on it rather than heading right for the target.
That's the root cause in my opinion. The details would fill a book...
As has been already mentioned, Artificial Intelligence: A Modern Approach by Rusell and Norvig (or AIMA) is essentially the only choice for serious study of AI. Your relative algorithmic naivite will make it a bit of a struggle, but there is a long history of smart physicists moving into AI.
Unfortunately, there is also a long history of smart outsiders getting trapped in "junk AI". These are the branches of AI that exist more because the metaphor is compelling rather that the results or prospects. These include: Neural Networks, Genetic Algorithms, Ant Colony Optimization, etc. I won't claim there is no good work in these areas, but there is too much fascination with the techniques themselves over the results, such that research constantly "solves" problems that would be done better with other techniques, but yet are somehow "interesting" because a neural net does it. The mainstream of AI is mystified why anyone would be interested in a technique that works 80% as well as the state of the art just because some guy in the 50s attached the word "neural" to it.
If you want to simulate brains, you should study neuroscience. If you want to know what's going on in mainstream AI, you should bone up on probability, statistics, and linear algebra (if you're the right kind of physicist, you already have the math you need).
Before you mod me as flamebait, please note that I do know what I'm talking about. My PhD is in AI and I'm professor in a CS department in an undergraduate engineering school, where I teach AI and Robotics. I was once the maintainer of the comp.ai FAQ, and I have published several papers in neural networks and genetic algorithms.
Re:Not quite as bad as it seems
on
Google Juice
·
· Score: 1
As far as googlewhacking is concerned, it's not as easy as it looks. Try 'parrhesia verboten'. I stopped once I found that one, proving to myself that it can be done.:)
I found it very easy. my first try was: 'monkey ukase' but I had about fifty hits. Then I tried 'bayes ukase' and got one hit. Now Bayes is really a proper name, and the page returned was just a word list, so I tried 'bayesian ukase'. Piece of cake.
P.S. Reverend Bayes discovered a lot of probability theory and has Bayes' Rule named after him. Bayesian is an adjective describing a mathematical approach as using probability, esp. Bayes Rule. A ukase is a proclamation that becomes law. It's from Russian via French and I usually hear it pronounced YOO-kase.
What I'm most intereseted in is not Lisp as a mainstream language, but rather Lisp as a research language. When people talk about the new things in programming languages these days, they talk about lazy evaluation, polymorphicly typed functional languages (e.g. Haskell). Since the ANSI spec, it seems as if Lisp has stagnated. CLOS gave us objects, but very little new has come down the pike since then. At one time, much of the new work in programing languages was done with Lisp. Now Lisp seems to be in the position of C: an excellent language that has aged out of the cutting edge. I guess my question is, is this a correct assessment, if so, should something be done about it and what should that be?
I agree with the sentiment here, but I already find pop-ups pretty annoying. I like banner ads, and I could never figure out why they were held to a higher standard than highway billboards or the sides of busses (try to click through on one of those!)
The good news in the article was that even banner ads have some effectiveness, especially at raising the viewers' consciousness about a company or product. Seems obvious to me, but maybe this means that people will start paying for them again.
I believe it is earnest, but that we're missing the point. They're calling for comments on their source for prior art of business method, when the real problem is the _very idea_ of a business method patent. Any comment made in response to the call that was on-topic would not adress the real issue: Business Method Patents are an anathema.
cheers,
ric
Call it crisis week de chinois. Hackers from a certain eastern nation threatened to attack US government sites over the course of a week. The IT folks a certain government academy fell all over themselves trying to prepare for an onslaught and in the process accidentally disrupted outside connectivity for a day and brought down the students' mail server for several days over finals. Of course, there were no attacks from the outside, but we sure had a week of crisis.
Operating System Definition
on
Is UNIX An OS?
·
· Score: 1
As has already been pointed out, the noun 'operating system' is a description of a certain class of program, and arguing whether FOO is an OS or not is really an argument about how we define 'operating system'. So the question becomes what is a good definition of OS?
Now I've taught operating systems at a major university several times, and I have a number of textbooks on the subject. All of them consider an OS to be pretty much a kernel plus a shell. Now Every would call me an "academic purist" and dismiss this definition, so lets look at how he defines an OS:
"An operating system is the software that comes with a computer (or OS distribution) that programmers and users need to make themselves productive."
Interesting...So if I need to relax in order to remain productive, and I relax by playing freecell, therefore freecell is part of the OS. Every essentially maintains that any software that is shipped with the kernel, no matter how irrelevant, is part of the OS.
Academics define categories to make useful distinctions between instances of larger categories. The currently accepted definitions of OS exist because people have found them useful. The only alternative Every have given us is clearly not useful: 'all software can be operating system, depending on how it's shipped'. Therefore I predict that most of us will continue to use the old definition until a serious replacement comes along.
cheers,
Ric Crabbe, Ph.D.
P.S. I don't consider the shell to be part of the OS, it's just another user program. But I acknowledge that's extreme.
Slashdot Mirror
These sorts of articles that pop up from time to time on slashdot are so frustrating to those of us who actually work in the field. We take an article written by someone who doesn't actually understand the field, about an contest that has always been no better than a publicity stunt*, which triggers a whole bunch of speculation by people who read Godel, Escher, Bach and think they understand what's going on.
The answer is simple. AI researchers haven't forgotten the end goal, and it's not some cynical ploy to advance an academic career. We stopped asking the big-AI question because we realized it was an inappropriate time to ask it. By analogy: These days physicists spend a lot of time thinking about the big central unify everything theory, and that's great. In 1700, that would have been the wrong question to ask- there were too many phenomenons that we didn't understand yet (energy, EM, etc). We realized 20 years ago that we were chasing ephemera and not making real progress, and redeployed our resources in ways to understand what the problem really was. It's too bad this doesn't fit our SciFi timetable, all we can do is apologize. And PLEASE do not mention any of that "singularity" BS.
I know, I know, -1 flamebait. Go ahead.
*Note I didn't say it was a publicity stunt, just that it was no better than one. Stuart Shieber at Harvard wrote an excellent dismantling of the idea 20 years ago.
[snip] ... "junk AI". These are the branches of AI that exist more because the metaphor is compelling rather that the results or prospects. These include: Neural Networks, Genetic Algorithms, ... [snip]
I suppose that your own articles written on 2nd order neural nets were part of your 'junky period' then, right?
Yep. And I paid for that too. While I was doing my PhD, someone at MIT was doing very similar work, but instead of using 2nd order NNs because they were cool, he had formulated his work with a solid mathematical base.
Guess whose dissertation was better received.
In my defense, I started in a time when the whole world was gaga over NNs and I was swept up in the hype. That's why I (like the ancient mariner) roam the earth issuing warnings to others.
Well, lots of reasons, the simplest being that it's a hard problem. But that's a cop out.
One issue we've had is that because intelligence is an observed phenomenon, not a defined one, its easy to think you're much closer to a solution than you are. The usual process is to observe intelligent behavior, and try to infer a formal problem from that to then try to solve. That problem eventually gets solved, and we discover we didn't ask the right question. Each failure has moved us closer in many important ways, just not directly at the target. It's a like a predator unsure of exactly where the prey is, circling and closing in on it rather than heading right for the target.
That's the root cause in my opinion. The details would fill a book...
As has been already mentioned, Artificial Intelligence: A Modern Approach by Rusell and Norvig (or AIMA) is essentially the only choice for serious study of AI. Your relative algorithmic naivite will make it a bit of a struggle, but there is a long history of smart physicists moving into AI.
Unfortunately, there is also a long history of smart outsiders getting trapped in "junk AI". These are the branches of AI that exist more because the metaphor is compelling rather that the results or prospects. These include: Neural Networks, Genetic Algorithms, Ant Colony Optimization, etc. I won't claim there is no good work in these areas, but there is too much fascination with the techniques themselves over the results, such that research constantly "solves" problems that would be done better with other techniques, but yet are somehow "interesting" because a neural net does it. The mainstream of AI is mystified why anyone would be interested in a technique that works 80% as well as the state of the art just because some guy in the 50s attached the word "neural" to it.
If you want to simulate brains, you should study neuroscience. If you want to know what's going on in mainstream AI, you should bone up on probability, statistics, and linear algebra (if you're the right kind of physicist, you already have the math you need).
Before you mod me as flamebait, please note that I do know what I'm talking about. My PhD is in AI and I'm professor in a CS department in an undergraduate engineering school, where I teach AI and Robotics. I was once the maintainer of the comp.ai FAQ, and I have published several papers in neural networks and genetic algorithms.
Parent should be modded up. Fluorescent flicker is a well known migraine trigger.
I thought everyone knew that Opportunity is a girl.
I found it very easy. my first try was: 'monkey ukase' but I had about fifty hits. Then I tried 'bayes ukase' and got one hit. Now Bayes is really a proper name, and the page returned was just a word list, so I tried 'bayesian ukase'. Piece of cake.
P.S. Reverend Bayes discovered a lot of probability theory and has Bayes' Rule named after him. Bayesian is an adjective describing a mathematical approach as using probability, esp. Bayes Rule. A ukase is a proclamation that becomes law. It's from Russian via French and I usually hear it pronounced YOO-kase.
Is this project:
cute? yes.
worth an A as a class project? yes.
Breakthrough? no.
Research? no.
New? no.
News? no.
-r
What I'm most intereseted in is not Lisp as a mainstream language, but rather Lisp as a research language. When people talk about the new things in programming languages these days, they talk about lazy evaluation, polymorphicly typed functional languages (e.g. Haskell). Since the ANSI spec, it seems as if Lisp has stagnated. CLOS gave us objects, but very little new has come down the pike since then. At one time, much of the new work in programing languages was done with Lisp. Now Lisp seems to be in the position of C: an excellent language that has aged out of the cutting edge. I guess my question is, is this a correct assessment, if so, should something be done about it and what should that be?
cheers,
ric
I agree with the sentiment here, but I already find pop-ups pretty annoying. I like banner ads, and I could never figure out why they were held to a higher standard than highway billboards or the sides of busses (try to click through on one of those!) The good news in the article was that even banner ads have some effectiveness, especially at raising the viewers' consciousness about a company or product. Seems obvious to me, but maybe this means that people will start paying for them again.
I believe it is earnest, but that we're missing the point. They're calling for comments on their source for prior art of business method, when the real problem is the _very idea_ of a business method patent. Any comment made in response to the call that was on-topic would not adress the real issue: Business Method Patents are an anathema. cheers, ric
Call it crisis week de chinois. Hackers from a certain eastern nation threatened to attack US government sites over the course of a week. The IT folks a certain government academy fell all over themselves trying to prepare for an onslaught and in the process accidentally disrupted outside connectivity for a day and brought down the students' mail server for several days over finals. Of course, there were no attacks from the outside, but we sure had a week of crisis.
Ob Ref to Jargon File: The problem seems to be with the implementation of the HCF instruction: http://www.tuxedo.org/~esr/jargon/html/entry/HCF.h tml
Now I've taught operating systems at a major university several times, and I have a number of textbooks on the subject. All of them consider an OS to be pretty much a kernel plus a shell. Now Every would call me an "academic purist" and dismiss this definition, so lets look at how he defines an OS:
"An operating system is the software that comes with a computer (or OS distribution) that programmers and users need to make themselves productive."
Interesting...So if I need to relax in order to remain productive, and I relax by playing freecell, therefore freecell is part of the OS. Every essentially maintains that any software that is shipped with the kernel, no matter how irrelevant, is part of the OS.
Academics define categories to make useful distinctions between instances of larger categories. The currently accepted definitions of OS exist because people have found them useful. The only alternative Every have given us is clearly not useful: 'all software can be operating system, depending on how it's shipped'. Therefore I predict that most of us will continue to use the old definition until a serious replacement comes along.
cheers,
Ric Crabbe, Ph.D.
P.S. I don't consider the shell to be part of the OS, it's just another user program. But I acknowledge that's extreme.