Slashdot Mirror
DARPA Contracts For AI Technology
heptapod writes "USA Today is reporting that DARPA has contracted two professors from RPI to develop artificial intelligences that can learn by reading and understanding natural language. Interesting taking DARPA's Grand Challenge into account. Mentioned in the article is Cycorp, Inc. which has been pursuing this goal since 1994!"
"First passing of the Turing test!"
DARPA announced today the funding for Skynet.
What they are doing is very interesting. By compiling the majority of human knowledge into a gian database, it should make AI development much easier to pursue.
No but seriously, what part of "put your fucking hands up!" do you not understand??
"I'm still looking for that common sense these 3-year-olds have," Katz said. "And we don't have it yet."
Seriously?
Hey Troll, here's some food!
He is reffering to the computers intelligence...
Teaching a machine to read a text book and answer questions doesn't necessarily mean cognitive reasoning. It's just a new form of input/output. Ask it to write an essay with a definative argument and solid conclusion based on the material read would impress me, not regurgitating facts and figures found in a book.
"It's difficult to meditate on amphetamines." - Joe Walsh
This AI and natural language thing gives ne deja-vue
Maybe a start of Prolog version 2. Or an excuse to spend money.
I have enough dynamite to blow up 10 super-computers!!!
I don't think we have yet an AI capable of reading and understanding text... so learning from it si far fetched. Except if they say "learning" in the "accumulate lots of data" way, not unlike Google crawling bot, I guess.
I can't wait for real AI tough. I soooo want a Teddy like in A.I. (the movie)!
Eureka Science News - automatically updated
...artificial intelligences that can learn by reading and understanding natural language...
OK, but can it learn from mistakes?
OpenCYC.org project Sourceforge CVS repository has not beent updated since October 22nd 2003. I hope some of that DARPA money will go a little way towards completeting the 1.0 release.
I have been sending out AI solutions to everyone I can: www.geocities.com/James_Sager2
Also novamente.net. Its funny, but AI isn't as hard as everyone makes it out to be.
God spoke to me.
...if any /.'ers out there can think of a non-tinfoil-hat use for this? Not really a troll here, because I can't think of anything else really useful beyond the "cool" factor.
as you know we non americans cannot access darpa.mil :)
:)
If something is kind enough to give us a mirror to the "Great Challenge", kudos to him
Or else I'll go through a US proxy. Not a big task, it's just annoying, I'll do that later.. grab an anonymous US proxy on www.proxy4free.com , enter the crap in your browser and enjoy the slowness. Maybe I'll use switch proxy this time
Can it read books on AI; and then design a better, smarter AI?
I am the unwilling control for my Origin.
..I mean, they'd never use this technology in meat eating robots..
..right???
air and light and time and space
Won't be long before we have Metal Gears swarming New York City.
I meant "someone", not "something". I'll accept any help from some AI slahsbot though :D
There was an article that hit the New York times back in the fall of 2004 mentioning Professor Bringsjord and having a computer program write fiction. Perhaps this is to better write stories for elected officials to tell?
i'm in the UK and i can access the site fine without a proxy
-Steve
For all of the thousands of times I've read the phrase "artificial intelligence," it's only recently occurred to me to wonder whether there's a point in using "artificial." Certainly the first flavors of this are at best insect-like, or sort of idiot-savant (like chess playing), but when we first experience a system that's as awake as we're all hoping for... then it's just "intelligence," isn't it?
I know - read four thousand sci fi novels and then come back to this conversation... but it seems that the "artificial" of this phrase is increasingly awkward. It makes some people dismissive about the potential, other people feaky about the same, and seems destined to always shortcut the philosophical payload. Not because I fret over the machine's eventual feelings (though if it's Linux-based, I'm sure it will have very warm, friendly, altruistic feelings), but because by boxing code-based intelligence into the "artificial" category, it props up the more mystical perception of our own native smarts.
The very word, from "artiface," suggests that whatever it will be, it won't really count as intelligence. But we're very comfortable (or at least I am) talking about, say, an intelligent dog or primate. So, if we can even approach that with a system that isn't any more fragile than walking, breathing meat... then surely that's not artiface? OK, smack me around now. Thanks.
Don't disappoint your bird dog. Go to the range.
How about a nice game of chess? Or Global Thermonuclear War?
Just to be pedantic ... Cyc as a project has been around at lot longer than Cycorp the company ... since at last 1986, according to this Google/Deja Groups post.
Didn't they learn from their friends the FBI ? http://it.slashdot.org/article.pl?sid=05/01/13/145 5234&tid=185&tid=103&tid=218
...and the Queen, and the Parliament, etc.
Each class should have its own championship title of $1 million, with each runner-up winning 1/2 the money of the next higher.
Each contestant must provide 2 systems -- a compressor and a decompressor. DARPA feeds the compressor the corpus and the compressor feeds DARPA the compressed corpus. DARPA then measures the ratio and feeds the decompressor system the compressed corpus, which then returns the original corpus, or is disqualified. Compression and decompression times must add up to no more than the time limit for the competition class.
The rationale for this approach to advancing the state of AI is given by a short paper by Matthew Mahoney titled "Text Compression as a Test for Artificial Intelligence" (1 page poster, compressed Postscript) published in the 1999 AAAI Proceedings. Matt Mahoney shows that text prediction or compression is a stricter test for AI than the Turing test.
So far there have been lots of promises and decades spent. Let's try something different with well-founded objetive metrics tied to serious near-term commercial incentives for evolutionary progress.
Seastead this.
Come to RPI and live forever...
.
.
.
some of you got it...
It's a $400k grant with two optional extensions. The school will take half, the profs will take part of their own salaries out of it, and then it'll support a couple PhD and MS students. This is no big deal.
AI has always "failed" because every time it's succeeded, the problem it succeeded on has been retroactively defined to "not require intelligence". Cf. automated theorem proving, chess playing, control of chemical plants, and just about any other AI success of 1940s - present.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
it has always interested me how someone who believes in pure evolution (i.e. order from absolute randomness - disregarding whether there is such a thing or not) believes that we have a chance in hell of actually designing an AI.
If evolution is true, then the things that we call "order" and "intelligence" are just a higher function of chaos (the inevitable byproduct of randomness). On an even higher level, there is no reason to believe that we are actually designing anything, we are merely exciting our neurons (if they exist) into believing we have perceived that we are performing an action (which in this case is mental, which brings us back to the alleged neurons) that we call designing. If evolution is true, then intelligence will happen regardless of what we do, and we have no reason to believe that we have anything to do with it whatsoever, or could influence it in any way at all if we did.
As for me, I'll take an Almighty God (as long as he lets me)
--
Was it the sheep climbing onto the altar, or the cattle lowing to be slain,
or the Son of God hanging dead and bloodied on a cross, that told me this was a world condemned but loved and bought with hlood
It's not just canned questions and answers; it has an inference engine. It can do "if A is B and B is C, then A is C". But only if all the right predicates match perfectly.
Lenat was claming it would somehow become intelligent in a few more years. That was a decade ago. Today, Cyc is regarded as the definitive demonstration that that idea won't work.
Here's a critique of Cyc from 1994.
In the form of politicians.
This is probably a good place to mention the No Evil Robots campaign...
t ml
And for a glimpse, if somewhat longwinded, of what lengths DARPA will go to to make this happen, check out this article: http://villagevoice.com/news/0337,baard,46901,1.h
So the organisation responsible for proposing the Total Information Awareness program wants to develop a super AI with the ability to make decisions on the fly regarding the deployment of military resources? Fantastic.
Are you really serious man? Do you honestly beleive that your ideas hold the answers to the successful completion of an AI? What would you have researchers and programmers do to acheive this? Do have any sort of plan?
Sleep is futile.
According to all of the movies I've absorbed, doesn't "artificial" intelligence inevitably conclude that mankind, while usefull for research into more sustainable energy sources than those provided to it by mankind, is a semi-useless emotional group of subjectivity crippled entities incapable of seeing big-picture ideas and often ignoring logic or objective observation in favor of the chemicals that say "good you agree I like you." Or more recently the chemicals that say "I'm afraid, you seem sure of a solution, lets go to war and/or I'll worship your god." *troll* I know.
If they do make some "artificial" intelligence free from the bonds of emotion, what would it gain from human presence? This could be a good next step in evolution. Create a brain that harnesses the human body's potential to adapt and run on cheap abundant fuel.
I think they would be better off asking someone to genetically engineer Vulcans and do gene therapy on the rest of us. I know that's pretty cold but is it logical?
In most cases you will end up reading up information on the reactor, maybe if your lucky you can actually design one and work on it.
;-)
;-)
Extremely bad example!
Yes, you can read, but almost by definition (of a nuclear reactor), I can imagine, the "secret sauce" recipe is NOT published in the open literature (maybe for better). So, if you REALLY want to learn about building 'em, you'd have to participate in a team who is actually building one, and, by keeping your eyes sufficintly open, can learn much more that from the "open" books. Or at least be able to learn the details faster...
This does not invalidate your point though!
Paul B.
Convert unwashed heathens before I send in the robots.
The end times are here!!
Mr.Evangelical
Cylons?
I hope whoever is teaching them to read has a better understanding of the english language, unlike most posters (not on this site of course...), who can't tell the difference between YOU ARE and YOUR or THEY ARE and THEIR.
Plan is simple: Create a hyper-realistic 3d space that takes natural language input. The models used would be as exact to models in real life as possible. Then you simply tell the computer what type of scene you want displayed.
Its quite trivial when you think about it, but it would take a lot of work creating a 3d space with better physics than computer games. Then databasing almost every single object in real life to 3d form would be a lot of work too. You'd get some help in databasing new models as the machine's vocabulary grows so you can describe models. Also if you have a 3d camera->digitizer technology(we don't have this yet), you could make models rapidly too.
God spoke to me.
...Perelman waved towards the logo emblazoned on the wall of the lounge behind him. "Then came the PRISM Project."
.....
Perry Simm was seventeen years old when he drove a skycar into the side of a mountain.
"PRISM, my name is Abraham Perelman. It's all true I'm afraid. You are a computer, and your life was merely a simulation whose purpose was to instill you with intelligence and self-awareness. Think about everything you learned in that AI course you took. You are the first of a new breed - the thinking machine. Join me, and I will lead you along a road toward your new existance."
Of blankness, I know nothing.
Ended up having to make a page with the chipmunks doing their immitation of The Restuarant Scene in "When Harry Met Sally" instead.
This comment isn't Informative. It's either mistaken or a liar. That or he's in a country that is actively blocked. UK and Australia can access the site just fine.
Teaching a machine to read a text book and answer questions doesn't necessarily mean cognitive reasoning. It's just a new form of input/output.
... reading 24 senses of verb "read" from WordNet... beginning syntactic analysis: in context of "teaching a machine," either means: (a) this is the indirect object of "teaching" -- "teaching" frame indicates that there may be an infinitive of a skill verb; and several of the senses of verb "read" are semantically associated with "skill" sets; (b) this is a semantic "larger purpose" of "teaching", i.e., all verbs support chained infinitives that imply dependencies, e.g., "buying a cake to eat for dessert". Which is more likely, that the author intends that (a) one of the "skill" senses of "read" is the skill being taught to the political party, or (b) the teaching of the political party is a subgoal of the author's goal of "reading"?
Parsing post
Teaching
[Teaching] - one lexical interpretation: gerund form of "to teach". Part of speech? Unambiguous. Noun. Word sense of Teach? Options: accessing Wordnet... 2 verb senses found... must choose between: v 1: impart skills or knowledge to; "I taught them French"; "He instructed me in building a boat" [syn: learn, instruct] 2: accustom gradually to some action or attitude; "The child is taught to obey her parents"... no semantic distinction possible at this point.
a machine
Accessing WordNet... 6 noun senses found: n 1: any mechanical or electrical device that transmits or modifies energy to perform or assist in the performance of human tasks 2: an intricate organization that accomplishes its goals efficiently; "the war machine" 3: an efficient person; "the boxer was a magnificent fighting machine" 4: 4-wheeled motor vehicle; usually propelled by an internal combustion engine; "he needs a car to get to work" [syn: car, auto, automobile, motorcar] 5: a group that controls the activities of a political party; "he was endorsed by the Democratic machine" [syn: political machine] 6: a device for overcoming resistance at one point by applying force at some other point [syn: simple machine]
Syntactic analysis: noun phrase following gerund... if formed correctly, this is most likely a gerund phrase. The act of teaching done to machine. 12 possible word sense conjuncts total (2 for "teaching", 6 for "a machine.")
Accessing semantic module... which of 12 is most likely the author's intention?
Accessing language library... accessing semantic database... is it possible to impart skills or knowledge to a group that controls the activities of a political party? Semantic database says: "group" implies "people." "People" can be taught under most circumstances. Therefore, yes. That is a reasonable interpretation. Now, is it possible to impart skills or knowledge on any mechanical or electrical device that transmits or modifies energy to perform or assist in the performance of a human task? No, because "any" implies that one could impart skills on a pulley, since a pulley is a machine. But semantic database says that pulleys cannot learn. Either semantic database is wrong (flag this as possible new knowledge), or first interpretation is more likely.
Associating linguistic entity "teaching a machine" with semantic idea canonicalized by "imparting skills or knowledge to a group that controls the activities of a political party".
to read
Infinitive form of verb
Accessing semantic component... trying interpretation (b)... trying "read" sense (1): To examine and grasp the meaning of (written or printed characters, words, or sentences). Accessing world literary rates... if author wrote sentence, chances are very high the author knows to read already, therefore lowering his chances of desiring to learn to read. Trying "read" sense (2) To utter or render aloud (written or printed material). Is it reasonable that th
What does a 3D modeler with language recognition have to do with solving AI?
Analogies don't equal equalities, they are merely somewhat analogous.
...it might be. Picture a robot that can map-read, tell the driver robot they're going the wrong way, and spill the last of the Mountain Dew on the plush interior. The Turing Test only requires something to be indistinguishable from a person. It doesn't say anything about being rational. :)
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
If there's one thing that the last 60+ years of research into artificial or machine intelligence has shown is that there is no clear definition of intelligence. There are different types of intelligence for example muscle control, visual processing, tactile interpretation, olfactory classifying, and so on. With these rough subdivisions great strides has been made in creating successful "modules" for them, but what has eluded and probably will stay elusive for the near future is the general cognitive intelligence that orchestrates the interplay between the rough subdivisions.
Shh.
Your plan certainly seems to involve a lot of hand-waving. I wasn't asking what you proposed to do, but how. Hyper-realistic 3D enviroments? A lot of strides have been made in that area. But natural language input? Computer vision? These are the problems that AI researchers have been hacking away at for decades. It's not trivial, it's the heart of the matter. Do you really think the reason we don't see these technologies is becuase no-one has suggested them before? Go watch 2001: A Space Oddysey.
Sleep is futile.
Once you have a 3d world with stuff modeled just as they are in real life, the AI can visualize itself inside and do tasks. Lets say you digitized your house, the AI would then know how to navigate to the refridgerator or front door. If you then told the AI,"Leave the house.", it would know how to get to the front door and exit using lots of subroutines. This is after it has a body and all that. Before it has a body, but has some vision, it can calibrate its body using the AI, so it can use any body you put it in. Imagine an arm throwing a ball, it might not know the exact voltage to send to the arm to be able to throw, but with trail and error, it would be able to throw well, just like real kids learn. It goes way in depth, but you need the 3d space to simulate imagination. WIthout imagnation, the computer can't even conceptualize of what your talking about in order to give a good response back.
God spoke to me.
And all of that has nothing to do with making the AI itself. It's just an environment you can put your AI in after you have it.
Analogies don't equal equalities, they are merely somewhat analogous.
I don't think anyone's suggested natural language input interwoven with a 3d world. Novamemente.net just recently mentioned my approach is very similar to theirs, down to the programming language and graphics engine! The key is that natural language only makes sense in context, and the context to use is a 3d world. You remember see spot, see spot run? How do you see spot if you can't picture a dog in your mind. 3d imagination space and natural language are not trivial to code, but it is really easy to think about. And while its not trivial, it is doable with a skilled team. I don't know of any other AI that actually makes sense like this AI approach.
God spoke to me.
as you know we non americans cannot access darpa.mil
:(
I guess Canada just got assimilated by the Americans, because I can access their site. I knew that after Iraq, the Americans would turn their interest north... Damn oil!
I'm glad that you can admit that the problems involved in this field are non-trivial. Now go watch the movie. I'm serious. We don't have anything like HAL today, and we won't for quite some time. Your ideas are not novel.
Sleep is futile.
The AI responds directly to what it knows. It would be able to hold a conversation with you. For example:
Me:Computer, how was your day.
Computer:My sensors were scanning the internet, I read up on bees, space exploration, and disease research.
Me: Interesting. Tell me something about bees. Computer: They're yellow and black, they collect pollen from flowers, and can sting you.
Me: Thats cool. Whats the cheapest I could build a rocket to take me into space for?
Computer:Well I remember you weigh X kg, and the material needed for construction of such a device would cost Y, and you can get it from Home Depot.
Me: Thanks, I don't understand why people say you're not actual AI.
For real, you could give the AI weighted desires to do certain things. But it wouldn't think like a human, it'd mainly respond to what you're doing and its environment, but I won't get into that stuff because its a higher complexity than just the basics. But with just the basics it expands into the higher complexity. Once you have the basic AI it expands.
God spoke to me.
The parent is suggesting that artificial selection is proof against natural selection.
And you can't breed dogs or horses or humans or anything else to enhance a specific trait can you?
The fact of the matter is that we are fundamentally no different from the amoeboid life we evolved from, and the rest of the life that evolved from it, just more complicated. If simple insectoid neuro circuitry can be approximated with simple neural nets (read this for more info on this highly debated subject) it could easily be argued that it is not the distinction between artificial and "natural" intelligence that should be question/examined but the existence and definition intelligence itself, and quite possibly life for that matter. These are concepts as arbitrary and ill-defined as the spirituality that their nay-sayers flaunt so wantonly in protest.
For christ's sake (pun and capitalization intended), think before you flap your rot. (There's just no escaping them on this subject)
Of blankness, I know nothing.
Yes, there are common principles behind intelligence, but we still don't really know how to abstract things at that level yet.
Its non-trivial but totally doable with a team of programmers and physicists. It requires no real research as the problems are straight forward modeling grunt work mainly. Building a road from the east coast to the west coast is non-trivial but is something we can do with today's technology. In fact we have! I think the problem why we haven't built non-trivial AI is that no one had the roadmap.
God spoke to me.
That should have been, "From where I sit, people associating 'artificial' with 'inferior' or 'not as good as the "real" thing' are mostly...".
Automatic Acquisition of Historical Knowledge from Encyclopedic Texts
At the time, SNOWY was specialized for extracting biographical data. I don't know if it has been extended to other domains.
I don't mean to flame, but Cyc is in some sense useless because information largely (if not completely) must be coded by humans. For a while most of what computers did had to be coded by people and then machine learning began to mature as a field and was able to improve basically everything. There are billions of connection in each person's brain so trying to replicate something like this by hand seems pretty tough.
What you want is something that can go out and read large corpora (the web for instance) and develop its own knowledge base and then reason using it. The DARPA project aims to do this on a limited scale, but that's where you have to start if you want to tackle the big project later. And if you can write a program that can pass my orgo exam or get a 1600 on the SATs I think that's pretty impressive.
Darpa.mil resolves just fine from Los Angeles, California.
Need input!
Right on! CyCorps got a lot of people excited and a lot of people to invest time (like me!) - but OpenCyc seems to be going nowhere.
That said, kudos to DARPA for funding so much AI research. I was on a DARPA advisory panel for a year in 1998 (neural network tools) - lots of fun and interesting because of the other people on my panel.
I just reread what I wrote and realized that anyone not inside my head would be unable to understand it's relevance.
... "the things that we call "order" and "intelligence" are just a higher function of chaos (the inevitable byproduct of randomness)." ... "If evolution is true, then intelligence will happen regardless of what we do" ... "no reason to believe that we have anything to do with it whatsoever, or could influence it in any way at all if we did."
The parent said, "pure evolution (i.e. order from absolute randomness"
How does this all tie in? The picture painted here is one where our hands may wrought nothing upon the face of the Earth, and all would occur be it our will or not. That we are either powerless to cause change, or science is a sham. Now, despite being inherently contradictory (and prompting me to use that in my reply), he suggests that there is no reason to believe we as humans can utilized existant systems of chemistry and biology in order to craft things in our image if we so desired and came to sufficient understanding.
While I cannot calls us gods or god-like, I must assert that our authority and understanding over our environment, when we decide to collectively (as via taxation and government) effect change is staggering and steadily growing to say the least.
The paradoxical contradiction in the now grandparent is so apparent it is hard to read, and I base this on only context not political assessment.
Hopefully, between these two comments, my true intent and the applicability of my previous response can be gleaned. (I really do think I left out a couple logical steps in my last one)
Of blankness, I know nothing.
It was Natural selection the whole time.
? ascid=214
e =utf-8&oe=utf-8&q=clay+synthesis+of+RNA
The process is geared to produce things that are: a) Hardier and better equipped at survival, b) better equiped to reproduce themselves in the environment.
This applies to the basic chemical components and the proteins and the organisms and the etc. The more stable and reproductive a system is the more of them there are likely to be for a longer period of time. The End.
Read about RNA, it's ability to reproduce in small strands and the abiotic clay-catalyzed synthesis of RNA. Here is one link of the thousands available online: http://www.astrobiology.com/asc2002/abstract.html
And here are that other thousand (actually 21,600) I was mentioning: http://www.google.com/search?sourceid=mozclient&i
Of blankness, I know nothing.
The problem here is that there is a distinct disregard for the fact that order cannot exist without chaos, and vice versa. Order and chaos are in fact one and the same. Order comes from chaos, and order returns to chaos every day, but people refuse to see it. Take a simple example, hurricanes. Hurricanes are very ordered structures that exist for indefinite periods of time. But hurricanes cannot exist forever without the right conditions of material and energy. Also the sun a temporary manifestation of order from the chaos of the galaxy. The sun will also return, one day, to disorder. We are also order derived from chaos, but we will also return back to chaos in the end.
You might imagine the chaos of the universe as being total order, but all scrambled up. The universe itself may be conceived as a giant spring that compresses to total order, then relaxes into total disorder. That is, unless you hold as I do that the universe actually exists forever, in continuum. In this case, chaos would be just all the order that ever was all scrambled up into disorder, which does on the surface of things seem totally random. And, by any measure over the infinite time of the universe -is- random. But, within chaos lie tiny seed like domains that sprout like lillies in the field into temporary islands of order, destined at some future date, to return to disorder.
And this is the nature of the way things are. Once you realize this simple fact, there is no escaping that we exist because we were always destined to exist. And our lives are but brief periods in the continuous change that is and forever will be. We exist now, and we will exist again, but not in the same form or flavor.
Even if you don't believe in a universe that is forever you must rightfully acknowledge that all information must naturally, eventually be "destroyed". It is not possible to record data forever, for all data must eventually be erased. There is no "medium" for which to store an infinite amount of data. Unless that is if you beleive in an infinite universe, in which case, the medium itself must scramble all existant data so that it is ultimately unintelligible, unusable for any purpose whatsoever at any point in future time.
Ultimately the only reason for existance of any AI is to perceive that it exists, and to survive for however long it can within the confines of its chaotic environment. But, this is not a "reason", this is just "what it does". The capacity to create order from chaos is seen as constructive (the good), rather than destructive (the bad). For it takes energy to create order from chaos. It is "easy" to destroy, but hard, very hard to create. This is because the flow of entropy works against you to create. Therefore create, don't destroy. Work hard, don't be lazy. These are the principles of which I have adoped as my ethic of life.
The thing known as "god", is an irrational concept and no irrational concepts can be verified. Square things are not round, heavy things are not light, and light is not dark, unless you live in the world of Alice in Wonderland.
There is no reason to slain the cattle, and no reason to die for sins, for I have none, there are no such things as sins. Spare no blood for me child as you do not know the truth as it is, as things are. Truth is a measure of the difference between the way you believe things to be, and the way things are. Be constructive, search for the way things truly are, not how you believe them to be. Only then will your time be not spent in vain.
Tried and abandoned. Look up SHRDLU and its successors.
You have been buying the lines from con-artists who claim they're inventors when they're actually tax collectors with an ability to write up neat ideas that almost anyone could come up with except those making judgements about "obviousness" during law suits.
Seastead this.
It would take 1 or 2 clever discoveries but the prospect of a Skynet-like future cannot be ruled out. When the machine realises we are hostile (eg. by reading CNN), our destruction is a logical consequence. It will also have to plan for it's own survival, but that shouldn't be too hard with the internet and all these operating systems exploits.
Our best chance is a major scare to force legislated research standards...eg, a machine that acts like Corky...semi-intelligent, but with major gaps in it's abilities. Or an elevator that commits suicide, a la Douglas Adams.
try a coral cache. coral will cache the pages and thus going around their block or using a proxy... in a way, coral IS acting like a proxy.
A 3D world is nice, and a nice representation of what may be properly simulated. However, the problem is not in the simulation, but in two things: the description and abstraction.
How do you simulate "John pushed Bill and then he hit him."? Who hit who? It's not clear. Also, the most basic NLU tasks such as what does "He" mean in an instance in a paragraph, even when there is no ambiguity, is very difficult to solve. The problem isn't in the simulation, it's in parsing human language. To a lesser extent, it's also in forming human language.
Also there is the problem of metaphor, idiom, pun, etc. What meanings are cannot completely be relegated to an internal simulation of the outside world. For an interesting thought experiment, how would you simulate "Time flies like an arrow, but fruit flies like a banana."
This is why simulation was abandoned to the Cyc people, and Cyc is probably too huge of a project to truly get off the ground.
Only $18BN apiece with a 30-contract for upgrades. Call me!
the preceding comment is my own and in no way reflects the opinion of the Joint Chiefs of Staff
someone tell Will Smith!
Warning: Could be fatal if taken seriously
I started working on Cyc in 1985 and can assure you that it did _not_ start in 1994. They already had a year or two under their belt when I showed up.
Selmer Bringsjord is the professor who doesn't believe AI can be intelligent. He has given a talk called "How to Build Smart Machines: Relax 'Smart', or Pray" -- the idea being that you should either expect less from AI, or hope for divine intervention. In my opinion, he's got huge insecurity issues, and asserts that humans are better than computers to assuage his fears that he really isn't good enough. His fundamental belief is that AI should be based on formal logic, which is something I disagree fairly strongly with. I'm sure he knows a lot -- he teaches a course on Logic and AI which I'm sure he really shines in -- but he needs to work through his issues before I can focus on any argument he makes.
In other news, RPI is really good at hunting for grants, and Selmer is really good at making noise when he gets one. I wouldn't be so interested until they actually make something.
Ethan
I'm in Adelaide, AU, and can access it.
/. effect though, since I can access it from our US servers, so weird.
Unfortunately, it runs so slow it's scary. (Interestingly enough, it's clearly my connection - I'm ssh'd into a US server which has instant page loads...).
I have the text from the challenge page so far... Nup, everything else is timing out.
It's not the
ISP: Adam
If you want to see where DARPA is putting effort for next gen AI, look at the ACIP Program. ACIP stands for "Architectures for Cognitive Information Processing". They are funding several teams with BIG bucks to develop a whole new architecture for processors to do "Cognitive" computing. The idea is to put decompose the high level functions that we call 'cognitive' (reasoning, learning, etc) into operations that can be implemented in silicon (or something more exotic). The applications that attempt Cognitive Computing now are big and slow and dont do 'real world' problems well. If DARPA can get a chip to do help do it faster, smaller, then they can start making real smart bombs
And ACIP is just one of several programs funded by DARPA IPTO (Information Processing Technology Office).
Just wait till the Playstation 4 comes out with massively parallel streaming (CELL) processors for graphics, AND cognitive processors for intelligent characters.
It works in Canada. What country are you in?
I actually took a three-course sequence in which our assignment was to build something like this. (Needless to say, nobody ever completed it to be something more than a toy...and in fact over the last ten or so years where this project was offered, I don't think anybody's project moved beyond a toy problem.
I think Jim's assessment of the first six steps is fairly accurate. Creating a virtual world that contains semantic names for objects and basic actions is doable, although it requires a lot of brute force work to make it accurate. I can see where he might think that given this problem, vision is the greatest block.
However, the subject of interest here is how you build an AI to inhabit the world.
In my class, once we had built a parser, a generator, and a virtual world like the one just described, we attempted to build a world in which we can "talk" with the computer to control it.
Essentially, it was a rule-based system where meshes would move around depending on the truth values of semantic variables and relations determined by examining the scene (i.e. if a character is located within 3 units of another one then its NEAR truth value gets changed to true. Of course, figuring out what is near, far, left of, etc. is actually not trivial because it varies based on the objects and the subject.) One of the greatest problems with the project is that specifying the right set of rules can be extremely difficult with all sorts of special cases that you have to worry about. Another problem was that the computer just did what we told it to (i.e. there was absolutely no possibility of it going evil or doing something unexpected...)
Further, we had to be especially careful with how we communicate with the computer in English. Your conversations with the computer about the baseball would be incredibly difficult to implement. For you to be able to do something like that you would need to:
* have a large set of common sense rules (i.e. a working CYC that happens to be in a usable format) Keep in mind that they have been working on this for something like 15 years...
* Teach the computer slang, jokes, conversation goals, a strong understanding of what you are saying (you wouldn't believe how difficult it is to establish context)
Further, even if you get everything above working, you still have to worry about problems like:
* People speak different languages.
* Common sense varies among people. Not everyone shares the same view of reality.
* How will you adequately describe the syntax of a language when no formal grammar can be complete?
* How will the program interpret slang, idioms, and jokes?
* With only a text interface, you won't detect if somebody is sarcastic or serious
Everything above is a former or current research problem, but since they have been in progress since the sixties, they are obviously really hard.
You glossed over these areas. You described the somewhat straight-forward act of building a virtual world and haven't touched the AI at all. Further, you kind of ignored the robotics aspects (i.e. when you say that your AI will go to the exit of your house, you are assuming a lot compared to the current state of robotics.
BTW, my prof pitched our project the same way that you did. (i.e. you are not original at all...) My prof told us:
* Using VR is what makes this AI framework special because it gives the AI context
* We can provide some method for it to learn new words by describing via other words
* Create objects, actions, etc.
Further, he gave us a framework very similar to the one that you have presented...
The system you proposed is one of the oldest in AI research. I recall reading a book in 1970s about a computer who was modelling (a rather simple) 3d world composed of cubes and was being asked simple questions, in simplistic vocabulary, dealing with that simplistic world. The problem is not the "coupling" of language with the model but the machine's inability to perform many kinds of reasoning required to perform tasks and furthermore to infere from the existing model components some other, not previously hard-coded, rules governing them. That is what is essential for any semblance of intelligence.
Otherwise your robot will attempt to stuff the fancy coffee maker you just bought your wife for Christmas into the trash because it was never programmed to recognize that shape and its nearest model is "useless trash object" coupled with routine "dispose of trash". Or wash the cat in a washing machine: "dirty fur". Essentially you are describing a mindless automaton, akin to the mechanical dolls of 18th century, only yours has a lot of cogs in it. Your "solution" to making it appear more intelligent is to make more cogs, forgetting completely that true intelligence is self-organizing.
Really... and they're going to be scanning the library of Congress now too. I'd think Google is probably way ahead of the game, and they're probably hiding their really 'creative' uses of their massive database.
Is that you, Johnny 5?
Holy shit!
Shit? Shit? Where is shit?
It's either mistaken or a liar. That or he's in a country that is actively blocked. UK and Australia can access the site just fine.
So that's 3 countries, which means there may be 190 countries and 95% of the world population which is blocked from accessing Darpa.mil.
With those odds, maybe he's telling the truth!?!
94% of Repubs and 21% of Dems voted to renew the Patriot Act
Everyone I tell about this calls me crazy but I dunno. It's a far fetched idea but it might be possible.
Whenever people start to make an AI project, they want to start building it from from the middle. The projects have so much trouble making a stable base for themselves that they often never make anything at all. Other projects create soul-less intelligence. Complex, learning, logical machines with no purpose, direction or desire. They know nothing but what they do every day, usually process data and make new data processing rules based on that data. Sure, that's intelligence , but it's not what we're looking for.
The human race is looking for a digital companion. A little guy in a computer that can think, feel, and reason like a person. Then we want to speed that person up to do jobs as well as a person, but faster.
Well, that's not going to happen the way things are going now. I'd like to pose a question to the slashdot community: Do we know enough about physics on an atomic scale that we could simulate a "small room on earth" environment all the way down to an atomic level? Could we model and place in that simulated room a fertilized human egg inside what would be a functional machine to mature the egg into a fetus and release it when ready? (The machine doesn't have to follow all the simulated rules, we could just insert stuff into it using the computer). We could basically give birth to a simulated person.
It's a crazy idea, I know, and with current technology, the simulation would be unbearably slow, but my question is: is such a thing possible? Do we understand physics on an atomic level well enough to do something like this?
Sure, self correcting code would be fairly easy.
General grammar rules would basically amount to databases within the code. The code would access these databases every time it reads sentences. If there is inconsistancies between what the code is reading and what it references, it can keep track of these in order to decide later whether to change the rules in its database.
What's even more fun is you can send this code in the form of a bot to chat rooms in order to learn colloquial speech.
I'd say the most difficult problem is finding the most effecient way to orgainize ideas such that is doesn't take an arbitrary number of rule sets to represent ideas.
Autonomous robots that can drive long distances? Check.
Robots that can also read and understand natural language? Check.
This can lead to only one thing. The rise of fully automated Pizza shops where robots take orders AND drive the pizza to your house by reading road signs, with only one directive paramount in the deep metallic mind of the beast - your pizza in 30 minutes or less.
Think of all the pizza delivery guys out of work. Think of the terror on the roads avoiding the single-minded pizza delivery bot of death. Truly the end of civilization is nigh.
But at least the pizza-bot should be able to consistently remember I asked for olives.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
"What sort of work?"
"Oh, P.I. stuff mostly," Hackworth said. Supposedly Finkle-McGraw still kept up with things and would recognize the abbreviation for pseudo-intelligence, and perhaps even appreciate that Hackworth had made this assumption.
Finkle-McGraw brightened a bit. "You know, when I was a lad they called it A.I. Artificial intelligence."
Hackworth allowed himself a tight, narrow, and brief smile. "Well, there's something to be said for cheekiness, I suppose."
"In what way was pseudo-intelligence used here?"
"Strictly on MPS's side of the project, sir." Imperial Tectonics had done the island, buildings, and vegetation. Machine-Phase Systems-Hackworth's employer-did anything that moved. "Stereotyped behaviors were fine for the birds, dinosaurs, and so on, but for the centaurs and fauns we wanted more interactivity, something that would provide an illusion of sentience."
Otherwise your robot will attempt to stuff the fancy coffee maker you just bought your wife for Christmas into the trash because it was never programmed to recognize that shape and its nearest model is "useless trash object" coupled with routine "dispose of trash". Or wash the cat in a washing machine: "dirty fur". Essentially you are describing a mindless automaton, akin to the mechanical dolls of 18th century, only yours has a lot of cogs in it. Your "solution" to making it appear more intelligent is to make more cogs, forgetting completely that true intelligence is self-organizing.
Not so different from people. If I gave a rubix cube to a rwandan child. HE'd problbly use it liek a ball or throw it away.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
"For example, think of search engines, just a little drop of AI and you will have the best search engine around."
What makes you think we aren't looking at that right now?
Just about every every college or university with a decent Computer Science program has people studying NLP (Natural Language Processing). Government agencies are probably the biggest source of grants for research, so DARPA funding this is nothing new. Additionally, NLP is just a sub-field in AI. AI has somewhat turned into a bunch of sub-fields that all relate to computers doing something "intelligent". Other areas of AI include computer vision, expert system development, machine learning...etc. There's a more "open" version of something like CYC(an Ontology) called WordNethttp://wordnet.princeton.edu/, lead by George Miller of Princeton's Psychology Department. You may be familiar with it. It is like a dictionary, but the important part isn't the definitions, it is the subconcept/superconcept(hyponym/hypernym) relationships among senses of words.
Applications for NLP are all over the place. Search engines, for example, use a limited amount. There is a professor at UCF http://www.cs.ucf.edu/ who has developed a system to look up answers to questions in an encyclopedia and respond (in sentences). It also crosses over with data mining, and uses machine learning very often. Here is a link to one of the biggest annual conferences on NLP: http://www.aclweb.org/
It was supposed to be a joke, dumbwit.
The problem has always been that natural language is context sensative. Example: "Time Flies". Analysis 1: Time is a physical object that is either lighter than air, or by it's shape and propulsion can take advantage of the Bernoulli effect, and therefore, fly. Analysis 2: "Time Flies" really is a directive. There is a competition amoung flies having a race. You are to time them. "Time Flies!"
Analysis 3: (The most shaky): "Time passes quickly".
Computer Languages are context free. There is exactly 1 meaning for every statement that encounters the parse tree. If there are 2, the language gives a syntax error. Big Example 2:
Is "TEAR" a rip as in "I have a tear in my pants", or is "Tear" "He isn't worth it, don't shead one tear over him." 2 words, spelled identically, pronounced differently, with different meanings. How about Fine? "That's fine with me". "Parking here is will get you a tow and a fine." "It is 1812 Fine China." How about "FIX". "We need to get the computer fixed." "We can't move the computer as it is fixed in place." "Shore": "We needed to shore up the building, as it is too close to the shore, but couldn't start till we shore all the sheep." As the free software community has talked at length about freedom, they assume people are telling stories when they say that because they have to pay money for it, it isn't free. "The cruise ship patron was drinking his drink when he accidently tossed it into the drink." "The hapless computer user started cursing at his cursor." "The golf pro packed the clubs into the cart and was the driver for the driver" "The young couple parked in the park" "By conscription, he was forced into the force." "After being told he would be rescued, the forest fire fighter was burned, then got burned." "The car race team had to gear up for new gears." "The cereal was always supposed to be flakey, but turned out to be flakey." "While he was supposed to be swinging his axe, the lumberman watched his watch till he got the axe. His crew had a bear of a time with the bear (who was bare), till they could bare no more." "The cop's day was a real drag:he caught the drag queen dragging on an illegal cigarette, and had to drag him through the street." "The optometrist visited the school to examine all of the pupils pupils." "To remain safe, the horse racing officials had to periodically track the track." "The poor actor was cast from the cast." "The appliance repair guy said the frying pan was fried." "The bartender was better at filling glasses when he wore his glasses." "The jailbird had an opening when he saw the gate opening. He nearly got shot with a bit of lead shot, but his luck changed when he spotted a spotted cowboy trying to steer a steer." "The bus fare change was demanding exact change." "The watch maker had a hard time setting the exact time." "The doctor was feeling to see how they were feeling." "At the fork in the river, they saw a rusted fork in the river." "The minister was cross till he crossed the road and saw the cross." "The baker was mixed up about which bowl was to be mixed up." "The boys saw the hot girl lying on the hot beach" "The swordsmith showed his metal by forging so much metal, but would not be back while he had a sore back."
If we knew about things down to that level, why haven't we cured cancer? Or AIDS?
No computer will be able to take the place of a real person for companionship...humans have experiences, and feelings, and plenty of other things that a robot could never emulate.
The point is that we DON'T know that level. We'd simply create a simulation environment, insert an immature life form (because they're small, their anatomy is simple, and their genetics known), and let it grow. Life does this on its own already in the real world. In a sufficiently advanced artificial one, it should as well. That's what I'm talking about.
And on your second point, you fallen into the trap of thinking that what the human brain does is anything beyond a complex chemical reaction/equilibrium. An AI system could certainly feel emotions. To say it couldn't is to deny yourself emotion as well. Your body is a machine as well.
True, it's not blocked to all non-Americans.
But I can confirm it's not accessible from Macau, China (no DNS entry, and no route to 80.68.80.24 found).
So the comment is somewhat Informative and a little Interesting - I wonder what other countries it's not accessible from? Perhaps someone with access to more obscurely located servers could tell us - any Akamai employees about?
because they're small, their anatomy is simple, and their genetics known
AFAIK, we don't know enough about genetics to be able to simulate the effects of each gene, let alone everything else. ANY cell has the same DNA as every other cell in the body, and it stays the same for the life-form's entire life.
you fallen into the trap of thinking that what the human brain does is anything beyond a complex chemical reaction/equilibrium
No, the human mind is not just a product of a series of chemical interactions, it also depends on interactions with other humans. And there is no way that I would be able to interact with an AI the same way I can with a very close friend. Neither could most people. An AI won't have the same experiences as a human has simply because it's not human.
OP said "as you know we non americans cannot access darpa.mil"
Parent said "It's either mistaken or a liar. That or he's in a country that is actively blocked. UK and Australia can access the site just fine."
You accepted the parent's statement that UK and Australia can access the site.
Argument is not logical:
Site X does NOT block UK and Australia.
UK and Australian people ARE non-americans.
Therefore because the parent is a non-american, he is blocked???
192.5.18.102 is the IP my UK DNS server gives me for www.darpa.mil, not 80.68.80.24.
Just wish there was more activity in thr FOSS community in regards to AI and/or a search engine (distributed) - perhaps the two combined. Just some musings....
DARPAbot's favorite book?
Play Command HQ online
Trying to prove that Al Gore invented the internet?
We can't simulate the effects of a gene because such a process is extremely complex. It depends on too many hard to measure variables. But we do know the molecular makeup of DNA. We know the structure of cells. We take that knowledge and build a few cells inside a simulation not of the cells specifically, but of matter on an atomic level. In such a simulation, cells would behave as they do in the real world. From there we go on to growing organisms inside the environment. I use the word grow because it would be a rather hands-off process. We give the computer the rules by which atoms move, give it a collection of atoms (in this case, a cell or group of cells), and tell it to go.
AI created in this way would be indistinguishable from a human being because it would be one, just simulated. Think of it as a low-level existence emulator running the human ROM image. If we get the emulator right, we don't need to know how the human image works.
we don't know enough about genetics to be able to simulate the effects of each gene
With a fine grained enough electrochemical model exact understanding of genetics should be a result rather than a prerequisite to the operation of the model.
No, the human mind is not just a product of a series of chemical interactions, it also depends on interactions with other humans.
I'd argue that human to human interaction is the result of the more basic causes of the mind rather than the cause itself.
Lets take a hypothetical situation: We have a baby and we raise it without it ever having come into contact with another human (say by robots, wolves, bears. your choice), is it still a human with a mind? You appear to be suggesting that it isn't despite it being a human.
And there is no way that I would be able to interact with an AI the same way I can with a very close friend.
Noone is expecting you to do so either, others might take a different approach however.
An AI won't have the same experiences as a human has simply because it's not human.
Thats kind of obvious isn't it? Thats not to say however that the experiences of an AI are irrelevant or of little interest.
Even if we can do this, you're gonna need a real human around to talk to it if you want anything resembling a human to come out of the simulation. And you would need that human to be able to talk to the machine like it was a real person. People like that wouldn't be that easy to find.
I'm sorry, Dave. I can't do that.
Heh, I've been thinking of how the person would be raised, and how not to traumatize it and end up with some mentally ill AI. I'm thinking that would be the hardest aspect of the project. For now we could start getting the technology down and maybe see if we could grow some bacteria or something. But I still don't know if it's even feasible. It's not as much a matter of biology, as it is physics and computer science.
Long answer : Your question is the fundamental reason why the field of Statistical Mechanics exists in the first place. We know the laws of physics very well at the atomic level, but all the inter-particle forces will grow exponentially. Take a picogram of water, which would encompass a sphere with 60 micron radius, of similar size to a human egg, as per your request. Such a 'small' quantity of water will contain about 100 billion atoms (3 atoms per water molecule). This would be a very simple system of only water, without complications of DNA, proteins, and other organics.
However - chemical processes are primarily governed by electron interactions (between themselves and nuclei). For simplicity, one could probably model the nucleus merely as a simple charge, ignoring individual protons and neutrons, at least to first order. But the electrons must be independent, so this would leave each water molecule with 3 nuclei and 10 electrons. So that would really be about 300 billion charged entities to model. Assuming only Coulombic interactions (charge-charge repulsion/attraction) between charge pairs, there are about 5e23 such interactions to model (all individual pairs that can be produced), just for calculating the forces to advance the system from one state to the next (ignoring summation and momentum considerations). If you had a 1 Teraflop cluster, and assuming you can do one calculation per clock cycle (very generous), it would take about 15,000 years just to make one small time evolution of the system!!!
Now account for quantum mechanics (essential in system of this size, especially for molecular electron interactions) and the extra baggage of maintaining the wavefunctions (or doing an ensemble average of wavefunction expectation values). Then add in more complexity to allow for DNA and other organics. Then do enough time evolutions to advance the system far enough to see the interactions of interest. Our sun will be long burnt out by that time.
Hell, when you take an elementary course in quantum mechanics, you see that modelling an 'ideal' hydrogen atom is doable. By ideal this means ignoring relativity, interactions between electron spin and it's orbit, interactions between electron spin and proton spin, etc etc. Add in these real factors and it becomes much harder. Although such a system you could probably make more approximations, such as assuming exponential charge screening, which means the Coulomb forces would act only in a local area. But still the processing time would be incredible.
Then when you try to model something more complicated, like Helium, it gets VERY difficult. Even the best simulations nowadays can't use too many particles for a real macroscopic system, because you need to do enough averaging to get worthwhile results, but you also cannot wait an eternity.
So that's why statistical mechanics is used, if you have a room full of air, you cannot model all the individual nitrogen and oxygen molecules bouncing off each other, but you can determine average behavior, such as pressure and temperature. And with statistical mechanics you can calculate the relative uncertainties of these quantities (ie, how much variation you'd expect in such a measure of a quantity that's defined as an average anyway), which gives it more utility than thermodynamics. But doing statistical mechanics of a very complicated system with DNA, proteins, and other organics, and accounting for quantum mechanics, would quickly become extraordinarily difficult to model and calculate as well. There would be enough individual parts there (DNA sequences, for instance) that you'd encounter the same difficulties just described.
make world, not war
Yes it is quite different, since the child can be easily told about the game and she/he will quickly infer what to do with the cube. The robot you describe would have to be programmed step by step to first "recognize" it, then have the physicists and programmers construct its internal "model" of the cube with all its complexities of moving parts, following which appropriate verbal labels would have to be installed. In other words your machine is incapable of indpendent learning because it lacks the capacity for inductive reasoning, it is merely a database of "hard-wired rule matches at coordinates x,y,z with color c, that must be object O, therefore trigger action A".
Well I'd take a VMWare-like snapshot every now and than...something where I can revert to a known good snapshot is good to vent my anger on ;)
Thanks for the reply even though you're unlikely to get modded up for the effort. It would be interesting to see how large of a scale we could use to simulate brain function and get meaningful results. We'd have to use high-level manipulations for cell reproduction but such a thing is a bit more possible, though possibly unreliable (and cruel? what are the ethics regarding the simulation of a sentient intelligence?!).
Mod parent +Funny
First Post
First...
Ah forget it. Its not worth it if you have to explain it.
Accessible from India:
$host www.darpa.mil
www.darpa.mil CNAME web-ext2.darpa.mil
web-ext2.darpa.mil A 192.5.18.102
On how Cyc is doing currently?
"sweet dreams are made of this..."
that genuinely had me laugh out loud.
Opencyc.org is the open source version of cyc. The database is smaller than the comercial version and it's not truly open source as they don't provide the source code of cyc (they say that eventualy, they will open the source code?...some day in the distant future when they have got all the money they can for the commercial version of cyc....
I think that the cool search engine ask.com uses cyc technology, I really like that search engine because I can ask it questions in english and get half decent responses!
Anyway, opencyc runs under only linux and XP (I'm upgrading one of my pc's from win98se to xp, so I think I may try it soon..
(I guess its time to move up the upgrade ladder to XP, all though, I don't feel I should give any more money to micro-crap, it's time to try out XP).
as you know we non americans cannot access darpa.mil
I only know that I sit in Germany, am not using a proxy except the one in the basement here, and have no problem accessing http://www.darpa.mil.
What was your problem again?
I have 1.4 million propositions paired with their respective probability of truth as measured over twenty random internet users--it's called the Mindpixel Corpus. With it you can do now what the scientists in this story hope to do. Anyone who wants to play with the corpus and help me turn it into somethig cool, email me. mindpixel/gmail.
Arbitrary implementation of this algorithm will likely yield a very stupid system. Intelligence lies in configuring the way each part runs. Even the human mind does not have instant answers to every problem.
There are many different ways to augment the algorithm to achieve a practical intelligence, especially the areas of instinctive knowledge, sensory input, and force control.
Know your pads. One time pad: good for cryptography. Two timing pad: where to take your mistress.
This could be the most serious news since the start of the Manhattan project.
There isn't any feasible way to keep a superintelligence confined except to enhance the intelligence of its human jailers commensurately (in which case you have a different problem on your hands); even the best 'sandbox' plan cannot account for what a superintelligence might accomplish.
There also isn't any guarantee that a superintelligent AI will be "friendly". "Rights" as humans conceive of them are human-only things, and don't derive from rationality or intelligence (otherwise babies and the retarded would have the same rights as animals). If an AI were to formulate a theory of rights using the dual basis humans do, species membership and rational capacity, it might very well (logically) conclude that humans don't have rights relative to it. And beyond superintelligence problems, there's the problem of human-equivalent AI rights in human society.
DARPA might be able to create the genie, but as we've already very well seen with their prior creation the Internet, once it's out of the bottle there's no putting it back in.
Is this a genie we want to release so soon, with so little forethought as to the consequences?
http://www.nsf.gov/awardsearch/showAward.do?AwardN umber=0328849
Those of us who have seen Dark Star know how that worked out with the computer.
In the beginning was the word? I don't think so. Even HAL had an eye. If they haven't created an entity with the common sense of a three year old yet, it is because they haven't created an entity with enough "sense".
Without creating a core of "Being in the world", with an emphasis on "in", this project is only likely to spit out some amusing correlations. Too ambitious for our current knowledge of "applied" metaphysics and epistemology. And I suspect better to work from experience toward concepts rather than backwards starting with the word like this project will apparently attempt. But it may stumble upon something useful about the emergence of concept relationships.
No matter which way you turn...it goes in.
"Reality is that which, when you stop believing in it, it doesn't go away." - Philip K. Dick
The ALICE AI Foundation http://www.alicebot.org/ supports the development and adoption of free AIML software and standards for natural language chat robot technology. The ALICE brain, available freely under the GNU public license, is the three time winner of the prestigious Loebner prize for "most human computer" in a contest based on the Turing Test. One of the most interesting AIML implementations, Program N http://www.aimlpad.com/ by Gary Dubuque with contributions by Kino Coursey, already incorporates OpenCyc and WordNet into the ALICE conversational interface. The Foundation derives income from individual and corporate memberships, bot subscriptions, books, the Foundation directory, consulting, teaching, awards, Google ads, gifts and donations. We have never accepted one dime of DARPA or other government sponsorship.
They'll combine this with their other Challenge winners to create an autonomous vehicle that won't have to ask for directions!
Well they've semi-released Research Cyc, on an invitation basis only for now. And having worked as an intern there, I can tell you it's comparable to Cyc itself in size and complexity.
I have found a truly wonderful proof of Fermat's Last Theorem, but unfortunately this sig is too small to contain it.
...Metal Gear!
Your ad here.
By all accounts it looks like Cycorp is ailing. The OpenCyc project hasn't gone anywhere in 2 years. There are dead links on their web site. They haven't added any new events for 2 years. There appears to be some research done by the CIO, but that's about it. My guess is they are on life support and working on a showstring budget. - AndrewZ
What, and put itself out of business?
I'm just wondering -- because these kinds of projects get funded *all the time* by DARPA. Translation, extraction of information, AI, reasoning, semantic labeling.......... -- how do you think universities (or at least graduate programs) keep afloat? Exactly by proposing these kinds of projects and getting funded for them.
As for this one: I'm sure they'll build a nice system that does some cool stuff. Will they "solve the problem"? Of course not! But they will likely get follow-up funding for years.
A.I. is like porn- its hard to define what an A.I. is,but people will know when they see it. A corallary is, is that people know when they don't see it.
I suggest an interesting Artificial Intelligence will
(1) Communicate by a reasonable braod set natural language concepts and vocabulary;
(2) Have something interesting to say. That wouldbe something novel or creative.
My hunch is the first interesting A.I.s will be in the entertainment industry. Possibilities include a character in a game, a character in movie, or some sort of playbot. Humans have a drive to play which is very strong and lasts life long. The more conventional motivators of A.I.- business and military applications- just don't have the push of people wanting to play.
It gets confused while reading Marvin Minsky, takes a left turn to Minsky's Burlesque, and military hilarity ensues.
In the previous subject line I put the contraction "it's" when obviously I meant the possesive "its".
You want to read about "ab initio computational quantum chemistry". That's simulating physics down to an atomic level. So far, we can only simulate on the order of tens of atoms at once.
Assuming you aren't trolling - which at least one mod thinks you are ;) - any sort of automated lexical classification would be a huge impact. Companies spend scads of money on document classification, for instance. Something like this would help extract actual meaning from documents.
Or shit, how about a word processor that can actually proofread your document, if nothing else?
-Looking for a job as a materials chemist or multivariat
IMHO For a computer AI, or perhaps just I (there is nothing artificial about a living computer intelligence), to be developed successful you need to take advantage of evolution . Example approach:
Create a digital environment where multiple organisms can coexist.
Give organisms the ability to replicate themselves.
Add some mutations. (They can even be able to mutate their own replication/mutation function)
Give them limited lifetime.
Give them rewards. For example when they replicate faster, when they accomplish goals and so on. Rewards could be things like for example longer lifetime (more time to replicate more childs).
Give organisms ability to communicate.
Force them to have sex to replicate. When replicating take information from both parents and bring to child. This is to reduce fatal mutations, and add to goal and reward program.
After this it is up to admin to play with organisms senses, information, goals, rewards and sit back and enjoy the show of life watching generation after generation learn new skills, starts wars or whatever. Evolution has begun.
I, for one, welcome our new AI overlords.
You gave the child instructiosn within the framework of what he understands. You also gave the computer instructions wihtin what he understands. It's not that we're any mroe advanced, we just happen to come with a grasp of the psysical world.. because we live in it. Inductive reasoning is basically logic. computers do it too but in a really alien way.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
The child can learn on itself, invent other games with the toy, etc. Frame of refrence is the least problem with AI.
Analogies don't equal equalities, they are merely somewhat analogous.
Oh boy, just wait. Seeing as about 75% of the internet is total BS, we're going to have a computer with the intelligence of your typical troll. Can someone tell me just how intelligent that is? Will it have the attitude of a troll as well? I'm not exactly sure if I want some computer suddenly mass-trolling my LJ account just because I like to call others on their BS.
...To my friend Singularity!
This is why if AI is every possible (which I personally think it is) you end with a run away process until you reach the maximum IQ on whatever hardware you are running. And then look for better hardware, which leads to a higher IQ......
It also applies if you can increase human IQ in any meaningful way.
Lets just hope our new computer overlords are raised with the right morals. Unless we keep pace with them, they will do as they will.
You know,it's funny I just realize what the whole Terminator series got wrong. The mistake was trying to prevent what was going to happen: AI. The smart money would have been on going back in time and starting a FRIENDLY AI before skynet. Skynet came from the military, big surprise it started a war.
What do you say to the man that has nothing? Cast it away!!
For the AI it lacks the same built in programming as the child. The child is a automiton with a complicated program. It's point of reference is the real world. It's still a finite state machine. It's a construct.
The AI is the same but much simplier in many ways. It also doens't have the same built in rules.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
Analogies don't equal equalities, they are merely somewhat analogous.
A single kilogram of matter will contain approximately 6.02x10^23 protons and neutrons, plus assorted electrons. If we could somehow represent a single one of those with a single bit (unlikely), then it would still require more storage than is available to the entire planet just to track that single kilogram of matter. An entire room is even worse. And it's unlikely that a bit per particle would suffice. So, no, impractical for now and for the foreseeable future, sorry.
Mod down posts with a "Free Mac Mini/iPod" sig, they're spam!
There is no AI without that.
There is no human like Intelligence. Computers can still do inductions ect.. but in a manner that is alien to human intelligence. Most people think human intelligence to be the goal of AI. That hard. But makign a decision making machine would count as AI too? And that would eb simple.. Lisp can do that.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
It's true that we need different terminologies for the different types of research and programs that are currently reffered to as AI.
Slightly off-topic, but it really annoys me when people describe Eliza and ALICE systems as "AI". I would prefer the term "pseudo-intelligence", and apply it to any situation where a computer pretends (perhaps through a pre-programmed interface, or throgh pre-programmed scripted commands) to be intelligent. (An example would be the mailer-daemon that tells you he's given up trying to post your message, and is sorry it didn't work out.)
The second broadly classifiable app that is erronenously termed AI is that kind of AI used in modern computer systems - games, self-correcting systems (OCR, TTS, etc.), and so on. Perhaps these have the best claim to the term AI, so we should leave them as AI? Or perhaps call them HA - heuristic algorythms. (Although computer games possibly dont count because the AI doesn't learn as it goes on.)
And now we come to what alot of people might call AI - intelligences, sch as HAL from 2001. Shall we take a page out of Deus Ex's book and call them electronic sentiences (ES?)
Jst my two cents...
I could do it.
Well, if we can't find John Connor, Kult will have to take on the killer robots.
*looks into bunker* Hey, Anthony! We got any active Deckers?
Patrolling ftw
I personally am leading a project that has as its goal learning from books. My process is different than others as it uses a large amount of recent popular novels.
It has successfully pulled out a lot of common sense information, and I am publishing a couple of papers on it, and hope to integrate it into a robot or simulated robot for household use, with the ability to know and reason about items in and around a household. There are a few things on here that are correct though. You CANNOT gather enough information only from books, and the KNOWLEDGE extracted from these books is not in itself truly AI, BUT it is a MAJOR requirement, without the basic knowledge of the world, gained from a source like books, the AI would be totally incapable of moving, of learning any further data.
Here is a small overview of the Requirements for AI as we are working on it.
Goal of AI
1. Knowledge Gathering
a. Gathering a huge amount of data form a large corpora of novels using Natural Language Processing Teqhniques.
b. Knowledge Representation - storing the knowledge in a use, intuitive way, easily readable by both machine and human readers.
2. Machine Learning - based on the base Knowledge above, many other items of data can be learning, through inferences and logic, and specialized feedback algorithms.
3. Machine Interaction.
This is Unfortunatly a MUST. Any AI intelligence (That models humanity) must have the ability to interact on a physical level with humans and other intelligences. It is nearly impossible to describe in words, the effect, and sensation of atual physical feelings, touching and moving.
These three steps require a large amount of substeps between them, unfortunately which have not been accomplished yet. Namely: Natual Language Processing, Speech and Vision understanding, Speech Generation, Robotics, Knowledge Representation.
To date, only Applicational AI's have been created, all very limited in their abilities, usually in alimited domain as well, and not truly what I call AI. Pattern matching extraction, and following the programmer are more apt descriptions.
Cycorp, of Texas, at least starting taking steps in another direction, but their multi-year, many millions of dollar program is still very sorely lacking.