Loading ...
Sorry, an error occurred while loading the content.

RE: [Artificial Intelligence Group] - Meaning modelling? was Turing

Expand Messages
  • Andre Clements
    thinking about nlp and symbolic modelling techniques the challenge is to develop a systems that does three important things - on the one hand (1)learning and
    Message 1 of 25 , Feb 12, 2005
    • 0 Attachment
      thinking about nlp and symbolic modelling techniques the challenge is to develop a systems that does three important things - on the
      one hand (1)learning and expanding knowledge of and by meaning modelling, constantly (2) exploring establishing and evaluating
      relations within the topography of meaning while (3) at the same time abstracting the meaning in a way that allows functional
      reduction of data - so it looks like the real challenge isn't so much cognition as meta-cognition. Does this make any sense!? What
      do the experienced A.I.ers think?

      A
      >


      [Non-text portions of this message have been removed]
    • Unmitigated Gall
      ... if you designed a learning system there is hope, even us humans ... confronted with unfamiliar slang, not to mention an unknown language Well my point is
      Message 2 of 25 , Feb 12, 2005
      • 0 Attachment
        --- In artificialintelligencegroup@yahoogroups.com, André Clements
        <aclements@i...> wrote:
        > if you're trying to program the responses - sure, you're lost, but
        if you designed a learning system there is hope, even us humans
        > have to learn how to deal with slang, and are at a loss when
        confronted with unfamiliar slang, not to mention an unknown language


        Well my point is that Turing was an idiot. Convincing a human that
        it is talking to another human is practically pointless. There are
        more important things we could be teaching machines than ebonics.

        It's like people talking about adding emotions to machines. Even if
        you could, it would probably be dangerous. The great thing about
        them is their predictability. Make them unreliable and you could
        cause serious trouble. I mean, what kind of moron wants to program
        PMS into a machine, other than an idiot or a masochist?

        You can make machines smarter, but you should make them emulate our
        strengths not our weaknesses.

        I suppose AI is about the fuzzy logic in a sense. Creativity,
        learning spontaniety. But is it really wise to make machines model
        human neurosis?

        > -----Original Message-----
        > From: Unmitigated Gall [mailto:Spammastergrand@a...]
        > Sent: 12 February 2005 10:40
        > To: artificialintelligencegroup@yahoogroups.com
        > Subject: Re: [Artificial Intelligence Group] - Turing
        >
        >
        >
        > --- In artificialintelligencegroup@yahoogroups.com, ARASH ARASH
        > <maziar_teh59@y...> wrote:
        > >
        > > hi all.
        > >
        > > i am from iran and new for your group.
        > >
        > > i am software engneering. but my english lang is not so good.
        > >
        > > i have many question about the philosophy of the mind and the
        > conflict it with some problems like "all nonmaterial philasophy
        > belive that the Mind process is not material process "
        > >
        > > but some of expriment -like Chines room or Turing test- all
        says
        > that mentall process can be impleament at futuer .
        > >
        > > and i think that the semantic section is only special for the
        > human and not machins.
        > >
        > > plz replay and response my question .
        > >
        > > sicerly.
        > >
        > > bye
        >
        >
        > The Turing test was a bit absurd to start with. You can luck out
        and
        > fool someone with basic greetings with a program. But someone
        trying
        > to determine if he was talking to a machine could guess pretty
        > quickly if he asked in an unusual way.
        >
        > With all the possible topics of conversation and forms of
        speech, no
        > one could program that much information or understanding into a
        > computer. Even if the computer did a web search for topics, it
        would
        > not know how to address questions it wasn't prepared for.
        >
        > How is Motorolla looking these days on the street?
        >
        > "Motorolla is a tech company involved in processors, wireless and
        > other electronic, communications equiptment."
        >
        > "But is the smart money on it?"
        >
        > "Smart is another word for intelligent. Money is a form of
        currency."
        >
        > "But would you bet the farm on it? Or do you like Biotech?"
        >
        > "Betting is illegal in most states. A farm is a place where
        > agricultural produce is grown. Biotech is an industry in which
        > technology is applied to biology and medicine."
        >
        > "What do you think of the new Green Day? And how about them
        Mets?"
        >
        >
        > A computer can not possible be programmed to respond to slang,
        > obscure cultural references even having the internet as a vast
        > library of resources.
        >
        > >
        > >
        > > ---------------------------------
        > > Do you Yahoo!?
        > > Yahoo! Search presents - Jib Jab's 'Second Term'
        > >
        > > [Non-text portions of this message have been removed]
        >
        >
        >
        >
        > Yahoo! Groups Sponsor
        > ADVERTISEMENT
        >
        >
        >
        >
        >
        > -------------------------------------------------------------------
        -----------
        > Yahoo! Groups Links
        >
        > a.. To visit your group on the web, go to:
        > http://groups.yahoo.com/group/artificialintelligencegroup/
        >
        > b.. To unsubscribe from this group, send an email to:
        > artificialintelligencegroup-unsubscribe@yahoogroups.com
        >
        > c.. Your use of Yahoo! Groups is subject to the Yahoo! Terms
        of Service.
        >
        >
        >
        >
        > [Non-text portions of this message have been removed]
      • Unmitigated Gall
        ... is to develop a systems that does three important things - on the ... modelling, constantly (2) exploring establishing and evaluating ... time abstracting
        Message 3 of 25 , Feb 12, 2005
        • 0 Attachment
          --- In artificialintelligencegroup@yahoogroups.com, Andre Clements
          <aclements@i...> wrote:
          > thinking about nlp and symbolic modelling techniques the challenge
          is to develop a systems that does three important things - on the
          > one hand (1)learning and expanding knowledge of and by meaning
          modelling, constantly (2) exploring establishing and evaluating
          > relations within the topography of meaning while (3) at the same
          time abstracting the meaning in a way that allows functional
          > reduction of data - so it looks like the real challenge isn't so
          much cognition as meta-cognition. Does this make any sense!? What
          > do the experienced A.I.ers think?
          >
          > A

          Modelling is interesting. I think understanding the limits and
          benefits of human vs machine skills is important. If we could make
          realistic models of human orgnas with cancer and the effects of many
          chemical substances, the computer could predict the effect of those
          substances. Machines only advantage over us is speed. They have no
          logic, reasoning. We have to build models and accept that they are
          merely sophisticated calculators. Never going much beyond number
          crunching.

          We evolved. And we are only as good as the obstacles we overcame.
          Creating artificial life that evolves involves making an environment
          as complex as the life interacting with it.

          Creating artificial intelligence has the enormous barrior of not
          even know how our own minds work. Synapses, neurons, chemicals.
          Adrenaline, oxytocin, nephonepherine. DNA, RNA.

          It's like asking a caveman to reproduce a computer. Our own thoughts
          are so enigmatic viewed from a materialistic or empirical point of
          view, it will be very hard to digitize information we dont really
          understand in the first place.

          Maybe neural networks and some of the life sciences will advance
          computer science by using computers to model what is there and
          figuring it out once its in a digital form.

          But lacking awareness, a machine is still a machine. Lacking
          willpower, desires, it can do nothing but what it is told to, other
          then a little programed random behavior.

          I was thinking of interests. Take ten subects. Ten programming
          models. Give them interests randomly as well as abilities numbered
          randomly, like those games where you assign a player 10 magic, 12
          speed, 5 strength, 8 intelligence.

          Then let the computer models grab 8 lines of information from a box
          titles, sociology, psychology, earth science, whatever.

          You have them 'mate' and the mates take a combination using 80% of
          their 'parents' 'genes' 20% randomly assigning interest or ability
          in other fields.

          Your're evolving knowlege bases, personalities of a sort. But your
          your telling it anything we dont know. Just to memorize a few bits
          of trivia about earth science. It may evolve in a sense, but is it
          really gaining positive mutation?

          And can it be taught to understand anything at all if it is not
          aware, conscious, alive? No.

          Programs cant understand. It is part of consciousness. And it might
          be some devine intervention type of thing where we simply can not
          make machines able to understand, because it is a facet of real
          life. And the Metaphysical concept of awareness.

          One of the greatest proofs of some kind of god is that it seems
          impossible to manufacture awareness.

          > >
          >
          >
          > [Non-text portions of this message have been removed]
        • Valdinei Freire da Silva
          ... Well my point is that Turing was an idiot. Convincing a human that it is talking to another human is practically pointless. There are more important things
          Message 4 of 25 , Feb 12, 2005
          • 0 Attachment
            >
            Well my point is that Turing was an idiot. Convincing a human that
            it is talking to another human is practically pointless. There are
            more important things we could be teaching machines than ebonics.
            >

            I do agree that Turing's Test is not the better test to AI, but saying that
            Turing was an idiot is too much.
            Think about the philosophical implications of a machine doing well in such
            test:
            1 - What is conciousness?
            It's very hard to define, but somehow everybody knows the meaning,
            at least theirselves.
            2 - Why everybody has a conciousness?
            Because I have, and once other humans is like me, I'm sure they also
            have. Few people ask themselves whether an ant has conciousness, but many
            more ask themselves whether an ape or a dog has conciousness, only because
            the last ones is more similar to human.
            3 - What does it mean if a machine do well in Turing's Test?
            First we would have to deny the arguments when answering question 2.
            Besides that, if a machine survive Turing's Test, can be sure it will show
            lots of sign of intelligence, for instance, learning. Turing only tried to
            design a more objective test, what I think is a great beginning, for who
            until today cannot yet define what is intelligence.

            >
            It's like people talking about adding emotions to machines. Even if
            you could, it would probably be dangerous. The great thing about
            them is their predictability. Make them unreliable and you could
            cause serious trouble. I mean, what kind of moron wants to program
            PMS into a machine, other than an idiot or a masochist?
            >
            What does means PMS?
            The predictability regarding a machine is the final result (reach an aim or
            maximize some amount), not the way to get there. Once we have good
            architeture for autonomous agents (machines), all we have to worry is how to
            define objectively the final result, not worrying about what the machines
            will do, it is even desired unpredictability, otherwise we wouldn't need
            intelligent and autonomous machine.

            >
            You can make machines smarter, but you should make them emulate our
            strengths not our weaknesses.
            >
            Maybe, what you call human "weaknesses" is only hardcoded program to avoid
            bad plans or even to overcome rational "slowness". So fear, pain, happiness,
            sociability, etc. could be consider as general plans, while rationality
            would help for specialized plans.

            >
            I suppose AI is about the fuzzy logic in a sense. Creativity,
            learning spontaniety. But is it really wise to make machines model
            human neurosis?
            >
            This means that you believe that the Turing's test would imply conciousness.
            Because is the only of machines having human neurosis.

            Well, these are my thoughts.

            Valdinei
          • Unmitigated Gall
            ... saying that ... Okay, so I tend to exagerate. ... in such ... meaning, at least theirselves. Is theirselves a real word? ... Except vegetables. And they
            Message 5 of 25 , Feb 13, 2005
            • 0 Attachment
              --- In artificialintelligencegroup@yahoogroups.com, "Valdinei Freire
              da Silva" <valdinei.silva@p...> wrote:
              > >
              > Well my point is that Turing was an idiot. Convincing a human that
              > it is talking to another human is practically pointless. There are
              > more important things we could be teaching machines than ebonics.
              > >
              >
              > I do agree that Turing's Test is not the better test to AI, but
              saying that
              > Turing was an idiot is too much.

              Okay, so I tend to exagerate.

              > Think about the philosophical implications of a machine doing well
              in such
              > test:
              > 1 - What is conciousness?
              > It's very hard to define, but somehow everybody knows the
              meaning, at least theirselves.

              Is theirselves a real word?

              > 2 - Why everybody has a conciousness?


              Except vegetables. And they could arguably be in a coma. Living in
              dreams.

              > Because I have, and once other humans is like me, I'm sure they
              also have. Few people ask themselves whether an ant has
              conciousness, but many more ask themselves whether an ape or a dog
              has conciousness, only because the last ones is more similar to
              human.

              Ants obviously do. You would have to get smaller.

              I saw a show on reproduction once. It showed sperm cells. When a
              woman had sex with another male, the sperm cells already there from
              another broke off into different tasks. Some went for the egg.
              Others formed a wall to stop invaders, still others stayed behind to
              attack the invading sperm of anothers DNA.

              What struck me is how purposefully they moved around. Like they
              could see. Amazing. Its as if they were conscious, aware, which begs
              the question, are all cells like this? Is our brain and body
              composed of billions of seperate consciousnesses?

              What does this imply? That the one is an illusion? I am many? The
              many come together to make me feel like one consciousness?


              > 3 - What does it mean if a machine do well in Turing's Test?

              That someone wasn't asking the right questions. Or the program got
              lucky? Anticipated basic greetings.

              Just by looking for ? at the end of an interogators wuestion, a
              program could respond "I don't know." Unless it was something it did
              know like its name, the time, etc.

              If the interogator ended a statement with a period, the program
              could say:

              "Yup." Or "I know what you mean."

              > First we would have to deny the arguments when answering
              question 2. Besides that, if a machine survive Turing's Test, can be
              sure it will show lots of sign of intelligence, for instance,
              learning. Turing only tried to design a more objective test, what I
              think is a great beginning, for who until today cannot yet define
              what is intelligence.

              A modern programmer would look at the test as absurd. Other than
              basic greetings and very casual introductions like, "Who are you"
              or "What's your name" a programmer could never anticipate the
              infinite number of questions, phrases, jargon an interogator couls
              throw at it. Furthermore, we don't know how to make a computer grasp
              anything. As I've said before, understanding is a facet of awareness
              and consciousness.

              >
              > >
              > >It's like people talking about adding emotions to machines. Even
              if you could, it would probably be dangerous. The great thing about
              > them is their predictability. Make them unreliable and you could
              > cause serious trouble. I mean, what kind of moron wants to program
              > PMS into a machine, other than an idiot or a masochist?
              > >
              > What does means PMS?

              Premenstral syndrome. Just a joke.

              > The predictability regarding a machine is the final result (reach
              an aim or maximize some amount), not the way to get there. Once we
              have good architeture for autonomous agents (machines), all we have
              to worry is how to define objectively the final result, not worrying
              about what the machines will do, it is even desired
              unpredictability, otherwise we wouldn't need intelligent and
              autonomous machine.

              That's true in the sense that a learning machine, one that could
              transcend a slave would need autonomy and unpredictability. But from
              a coders point of view it is difficult. The idea of programming is
              to give a machine instructions. To tell it what to do.

              >
              > >> You can make machines smarter, but you should make them emulate
              our strengths not our weaknesses.
              > >
              > Maybe, what you call human "weaknesses" is only hardcoded program
              to avoid bad plans or even to overcome rational "slowness". So fear,
              pain, happiness, sociability, etc. could be consider as general
              plans, while rationality would help for specialized plans.
              >

              You can't make a program feel because it is not aware or conscious.
              It is dead. A program is like an algebraic statement, proof or
              formula. How do you make an algebraic formula that is aware of it's
              surrounding? That can feel, or think, or understand. That is what
              programmers are dealing with. It is like asking an authors
              characters to answer questions. They are not actually alive. Giving
              instructions to a machine is like pushing its buttons. But it
              doesn't know it has buttons because it is no more aware than a
              pencil. It is only a tool. We can try to emulate life. But we can
              not make code aware of its surroundings.

              We can emulate sight in a sense. Or hearing. But a computer can
              neither see or hear. Only take input.

              > >
              > >I suppose AI is about the fuzzy logic in a sense. Creativity,
              > learning spontaniety. But is it really wise to make machines model
              > human neurosis?
              > >
              > This means that you believe that the Turing's test would imply
              conciousness. Because is the only of machines having human neurosis.
              >
              > Well, these are my thoughts.
              >
              > Valdinei

              No, I don't think Turings test would imply consciousness as much as
              it would require consciousness to grasp the meaning of any word in
              the English language.

              Hey, they have things like computer psychologists. Interesting. And
              in a very real sense, comething that achieves Turings goal. But only
              in a very narrow predictable way. You could fullfill his criteria if
              the field and interaction was narrow enough. But there is no machine
              in the world that I couldnt determine was a machine in under 10
              questions.

              You simply ask it very human things. Not predictable things like
              whats your name, what time is it, how far is Saturn from Jupiter.

              You would ask,

              "Do you think David Gilmore is a cool guitarist?"

              "Isn't Terry Hatcher a hottie"

              You would ask it idioms. Questions without literal translation. I've
              done a lot of writing. Fiction. Studied speech, jargon, lingo, slang.

              I would guess 20% of our statements would be illogical translated
              literally.

              Go to another usergroup. And really examine phrases:

              "My bio test was a motherfucker."

              Can a biology test have incestuous relationships?

              A lot of human speech defies literal translation.

              "I aced that test. I was all over it."

              "You rock. But I have to grab some shut eye."

              "No shit."

              "Shit."

              "Okay, later dude."
            • cic@ua.fm
              http://www.directgalleries.com/?r=admincic http://www.directgalleries.com/?r=admincic
              Message 6 of 25 , Feb 13, 2005
              • 0 Attachment
              Your message has been successfully submitted and would be delivered to recipients shortly.