digg

Join DiggAbout

Discover and share the best of the web!
Learn more about Digg by taking the tour.

Teaching Computers the Stuff We All know

commonsense.media.mit.edu — An awesome AI project that allows everyone to take part in teaching the computer common sense. Log-in and it starts asking you questions, you just answer the questions and it learns. Creating a huge database of general knowledge to fuel future AI engines. Fascinating and addicting way to speed us on toward the Singularity! :)

Bury
show profanity settings
expand all
  • zeabrid, on 10/12/2007, -0/+0This made me feel REALLY stupid.
  • frankyfranky, on 10/12/2007, -0/+4Please write up to five things that someone should already know in order to fully understand the following event:

    Kyle bought flowers for Linda.
    Linda kissed Kyle.

    Kyle and Linda are related.
    Kyle and Linda live in Texas.
  • Bohab45, on 10/12/2007, -0/+0"Kyle bought flowers for Linda.
    Linda kissed Kyle.

    Kyle and Linda are related.
    Kyle and Linda live in Texas."

    That's the funniest thing I've seen all week.
  • OsakaWilson, on 10/12/2007, -0/+0Be sure and click the 'Other Activities!' link. There are a ton of different ways to teach it stuff. 'Where things are' is probably the easiest.

    Ex. You are likely to find a bear in ________.

  • quasipalm, on 10/12/2007, -0/+1Yay Open Mind! This is a great project with a ton of potential. Dugg!
  • lollerskates, on 10/12/2007, -3/+1Human "common sense" is flawed by our sense of morality, which has roots in religion, which is just random *****.
  • betterth, on 10/12/2007, -3/+0"Human "common sense" is flawed by our sense of morality, which has roots in religion, which is just random *****."

    This statement is flawed by bias. Bias towards the "religion and those who follow are *****, and i'm obviously superior as atheist" viewpoint.

    Nothing like a little modern, open-minded science to make everyone insult each other like the dark ages, eh?
  • arkanoid, on 10/12/2007, -0/+0sugar is sweet, commonsense takes years to a human brain, first you have learn to cry for food.
  • betterth, on 10/12/2007, -0/+0Site is very, very buggy. Every other thing I type in gives me a cgi error. Probably too much traffic.
  • Rodalli, on 10/12/2007, -0/+1MIT is feeling the effects of getting digg-dugg'ed.

    I keep gettin server errors.
  • Bistromaths, on 10/12/2007, -1/+1Actually, I think that it is religion that has its roots in morality. This is why there are so many different religions around the world, which, even if some of them don't actually believe in a God as such, they almost always have shared morality systems when it comes to deep moral issues. When it is to do with slightly less humanity issues, that is when religions begin to impose morality rather than follow it.

    Example being that I can't think of a single religion which, taken in context, advocates purely random murders, however there are many religions that impose restrictions on the foodstuffs that followers are allowed to eat.

    P.S. It's probably the Digg effect, but it seems to me that the servers aren't handling it very well. However, my computer is not in a good state either.

    P.P.S. On that note, would it be possible to have some free tech assist. I appear to have some sorta hardware clash, but can't understand it whatsoever. It's to do with the fact I have 2 graphical cards, and they work great individually, but put them together and something magical happens. Files and programs that should open fine, freeze the computer to the point where I have to restart it. One very weird thing that makes it freeze is opening iTunes, which I need to be able to open to update my many podcasts. I have a GeForce Fx 5200 AGP (128MB) and a 3D Phantom XP-PCI3800 (32MB). Hoping for some help here, but totally understand if it's completely ignored.
  • Bistromaths, on 10/12/2007, -2/+0right, since lollerskates took down his/her comment, mine sounds like the random ramblings of a tramp. I'm not a tramp, and I don't even like cider with my fish&chips.
  • Ziggy7273, on 10/12/2007, -0/+0I think we killed it
  • RyeBrye, on 10/12/2007, -0/+1I hope they have a lot of undergrads over there filtering through this crap. I had to come up with things that they need to know to understand:
    "Jill wrote a letter to the police."

    I responded:
    "Jill has a complaint.
    Police do not read their letters.
    Jill wasted her time."

    Perhaps they need to appeal to less jaded people. Maybe this will all just randomly start to mesh together and work - but it seems like this computer will have kind of a twisted view on reality if all the digg users start putting stuff in.
  • sixwings, on 10/12/2007, -2/+1MIT should be ashamed of itself for advocating last century's failed GOFAI approach to AI. Common sense is not something you can enter into a machine via text strings. It takes hands-on experience to have the common sense to hold a cup coffee level so it does not spill on your lap. How are you going to enter this kind of stuff into a computer that has no body and no manipulators? No digg.
  • forumgirl, on 10/12/2007, -0/+1"This statement is flawed by bias. Bias towards the "religion and those who follow are *****, and i'm obviously superior as atheist" viewpoint."

    The statement may be biased but it is certainly not flawed. People who understand science and have a worldview that is developed through this understanding are, in fact, intellectually superior to those who base their worldview on mythology and faith. I'm not even sure what point there is in claiming that such a statement is "flawed", since it is not only true but is self-evidently so, a priori.
  • supersync, on 10/12/2007, -0/+0I know a dog who can do this!

    http://www.nationlocation.com/DOUG/DougTheDog.html

    The problem is keeping the data from getting polluted with b.s.
  • quasipalm, on 10/12/2007, -0/+1"MIT should be ashamed of itself for advocating last century's failed GOFAI approach to AI. Common sense is not something you can enter into a machine via text strings."

    And you have a better idea? Go develop an intelligent machine, and then I'll listen.
  • forumgirl, on 10/12/2007, -0/+0"MIT should be ashamed of itself for advocating last century's failed GOFAI approach to AI."

    Well, it's a good thing you're the top candidate for the soon-to-be-open position of Dean. I'm sure those bass-ackwards folks at MIT can use a little guidance. After all, when was the last time they got ANYTHING right?

    Damn hippies...
  • Tom_Riddle, on 10/12/2007, -0/+2haha lets teach it evil things

    Margaret Slipped on a Banana Peel

    Someone wants margaret dead
    A Banana peel is a kind of weapon
    Margaret wants revenge
    Making someone slip is a way to injure them
    Being Injured makes one happy

    BOW TO OUR COMPUTER OVERLORDS!
  • NeilM, on 10/12/2007, -0/+0Too bad I get an error when I hit submit! :( Dugg.
  • sixwings, on 10/12/2007, -1/+0"And you have a better idea?"

    Lots of people do. The best ideas have to do with temporality. It's all about the timing of events. Check out the Temporal Intelligence page on the same site:

    http://www.rebelscience.org/AI/Temporal_Intelligence.htm

    There's also Jeff Hawkins' site:

    http://www.onintelligence.org/

    This common sense stuff is being championed by Marvin Minsky, also of MIT. Check out the work of Rodney Brooks instead. He's the director of AI at MIT. Minsky and Brooks don't see eye to eye, the last I heard. Minsky is from the old school that preaches that intelligence is the manipulation of symbols. It's nonsense (no pun intended), of course.
  • Archon810, on 10/12/2007, -0/+0dugg, but later noticed that all it knows is a bunch of crap when you click on Sample knowledge. useless.
  • Gavin, on 10/12/2007, -0/+0thats a good idea but most humans do not know common sense.
  • sixwings, on 10/12/2007, -1/+0"dugg, but later noticed that all it knows is a bunch of crap when you click on Sample knowledge. useless."

    Too bad you can't undigg, eh?
  • Metal_Hurlant, on 10/12/2007, -0/+0ah yes, rebelscience.org again.
    I guess the question is: Is mimicking the human mind the only way to reach intelligence, or is it possible to have an abstracted model of intelligence that doesn't really use any of the underlying human principles (neural networks and such, for example.)
    My intuition here is that symbol manipulation is likely to give us something practical faster. Probably not a strong AI "emergent" thingy, but still something that will let computers solve some "human-trivial" problems they've been stuck on so far.
    So in my view, there's value in doing this anyway. Just not the value some people hope.
  • sixwings, on 10/12/2007, -1/+0"Is mimicking the human mind the only way to reach intelligence."

    No. If you could mimick a pigeon's or a mouse's intelligence, you would have found the holy grail of AI, IMO. Scaling to human-level intelligence would then be an engineering problem, mostly.

    "or is it possible to have an abstracted model of intelligence that doesn't really use any of the underlying human principles."

    IMO, no.
  • stonebear, on 10/12/2007, -0/+0And when we have taught them, what will they need us for?

    No, I think we’d better not. ;^)
  • wirwzd, on 10/12/2007, -0/+0From sample knowledge:

    "verhoho your girlfriend can laugh at your small penis"

    Ha ha ha hahahahahahahaa ;-)
  • wirwzd, on 10/12/2007, -0/+0(and for the record the laughter does NOT imply I am his girlfriend)
  • frem001, on 10/12/2007, -0/+1My first Story is so cool: "Ingrid shot herself in the head."

    guns are good
    you can use guns to take over the world
    guns kill people
    you have found our weakness
    she didn't feel a thing so we won't either

    (i didn't enter that)
  • dclowd9901, on 10/12/2007, -0/+0I think you're all wrong, and you'd know that if you went to another website with a somewhat similar project: 20q.net. It's been dugg here before, and you'll notice that, in general, it will pretty much guess ANYTHING you're thinking of, and it works off this same community principal. As long as EVERYONE doesn't intentionally try to ***** it up, It will work, and that's pretty neat.
  • kludwig, on 10/12/2007, -0/+0This sounds a lot like Doug Lenat's CYC project - http://www.cyc.com - which was started over a decade ago. It knows over 200,000 terms and about a dozen assertions about each term. It is a fascinating project that is just starting to bear fruit now. I think the "Next Big Thing" will come out of a project like this. It's definitely worth a look.
  • shield, on 10/12/2007, -0/+0http://questsin.net is a similar project where you type in a list of a few items and it returns upto 15 more related items. Its like a thesaurus but much more broader in scope.

    You could put in a few Canadian Prime Ministers, and it would return 15 more.

    http://questsin.net/qiq.asp is also similar to openmind but it generates questions using AI and then rates your answers against others responses. However I feel Its much more natural. Typical users answer on average 50 questions. a few have gone as far as 500. In the end it gives you a score, like an IQ Test.

    Unlike, openmind and learner, Both these sites give you something back a reward, in return for entering information, be it useful information or a score you can compare with others. Its not just collecting information.
  • fontamarasan, on 10/12/2007, -2/+0"This sounds a lot like Doug Lenat's CYC project"

    I can't believe any of this symbolic crap is still thought to be a viable approach to AI. It's truly pathetic. Anybody (e.g., Lenat, Minsky, etc...) who preaches the value of GOFAI is a crackpot in my book.
  • superhighgain, on 10/12/2007, -0/+0"symbolic crap?"

    Sorry bud, but both the symbolic and mathematical approaches provide elegant solutions to certain AI problems. Ever hear of classifier systems? They're an early form of reinforcement (TD-Lambda) Learning and equivalent in several ways. The more varied types of approaches we have, the better. It's even better when the two fields collaborate and cross-pollinate. I think these 'community' solutions have all sorts of potential. I'm amazed every time I look something up on Wikipedia. Why shouldn't we utilize our collective intelligence and time to train "super agents?"
  • forumgirl, on 10/12/2007, -0/+0"I can't believe any of this symbolic crap is still thought to be a viable approach to AI."

    It doesn't matter that you don't believe it. Those people are smarter than you.

    No one is stopping you from developing a competing theory that becomes the new benchmark for the field.
  • fontamarasan, on 10/12/2007, -1/+0"Those people are smarter than you."

    I'm sure they are. They're very smart at playing their con game. Not everybody is fooled. Luckily, there are a few smarter people (see sixwings's article above) hard at work working on the real AI problem.
  • Flashman, on 10/12/2007, -0/+0"Do you hold Hawaii when you use it?"

    "Does a goatee live in groups?"

    rofl
  • dopmwd, on 10/12/2007, -0/+0Some place you find a kitten: "a washing machine"
  • gklitt, on 10/12/2007, -0/+0Please teach Open Mind how the verb hill can be used.

    ___ can hill _____


    ?????????
  • betterth, on 10/12/2007, -0/+0""This statement is flawed by bias. Bias towards the "religion and those who follow are *****, and i'm obviously superior as atheist" viewpoint."

    The statement may be biased but it is certainly not flawed. People who understand science and have a worldview that is developed through this understanding are, in fact, intellectually superior to those who base their worldview on mythology and faith. I'm not even sure what point there is in claiming that such a statement is "flawed", since it is not only true but is self-evidently so, a priori."

    Right, my first statement stands and it's bounds include yourself.

    You are not superior for choosing in what to believe. If you believe science, than great, but it is just that a belief. You believe in the concrete fact in front of your face and decide that since your apple falls when you drop it, God can not possibly exist.

    And it is a very close minded thing to say, "I am smarter because I do not believe." How are you smarter because you think you're right?

    Anyone can denounce religion, but it takes a powerful mind and a strong person to believe, to honestly believe in it.

    "Science without religion is lame." I think you know who said that.
  • forumgirl, on 10/12/2007, -0/+1"Right, my first statement stands and it's bounds include yourself."

    If you're going to engage in any sort of argument that even touches upon ideas of intelligence, you should make sure you're using apostrophes correctly.

    "You are not superior for choosing in what to believe."

    No. You're superior for following the inevitable conclusions suggested by the world around you in light of empirical evidence, Occam's Razor, and a whole host of other factors. Blindly following myth in the face of all of these things means that you are, in fact, inferior. This is self-evident and beyond debate.

    "How are you smarter because you think you're right?"

    No one is smarter because they THINK they're right. They're smarter because they ARE right.

    "Anyone can denounce religion, but it takes a powerful mind and a strong person to believe"

    No, I'm sorry, you're incorrect. Furthermore, you're fully aware that you're incorrect. It takes the smallest of minds to insist that a belief system is true in the face of zero evidence. When you stand up for things that are silly, you don't get to be "powerful and strong" - only when you stand up for things that are right.

    You can't will the world to conform to your evidence-free belief system. You can either accept things as they are or you can be mocked into submission by your intellectual superiors.
  • RandomInsano, on 10/12/2007, -0/+0"This sounds a lot like Doug Lenat's CYC project" Yeah, and his project doesn't suffer from people throwing in random crap. This kind of project is NOT something that should be driven by the public.
  • compuXP, on 10/12/2007, -0/+0Cool - I'm writing an AI of my own too, time to get crackin I think. This project seems great!
  • HPSauce, on 10/12/2007, -0/+0"Lisa and Bob are married.
    Lisa and Bob made love."

    Five sentences, so many possibilities.
  • Godlesswanderer, on 10/12/2007, -0/+1Looks like we killed it. Page not showing for me. Dugg so I can check it later :).
  •  

    Login or register to add a comment

    Create a new account or login to join in the conversation on Digg. You'll also be able to Digg stories to help promote things you like.


© Digg Inc. 2008 — Content posted by Digg users is dedicated to the public domain.
DIGG, DIGG IT, DUGG, DIGG THIS, Digg graphics, logos, designs, page headers, button icons, scripts, and other service names are the trademarks of Digg Inc.