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

Re: Sim vs non-sim (was sim and copy)

Expand Messages
  • John J. Gagne
    Happy New Year to all the group! Hi JC: I think your post below was quite good! ... That s one way to look at it. A good way as for our purposes. Mostly, I
    Message 1 of 8 , Jan 1, 2008
    • 0 Attachment
      Happy New Year to all the group!

      Hi JC:

      I think your post below was quite good!

      >
      > You cannot simulate intelligent behavior, it is
      > either intelligent behavior according to some
      > criteria or it isn't.
      >

      That's one way to look at it. A good way as for our purposes.

      Mostly, I like your reference to stating the criteria. But it's just
      that reference which could be used to falsify the sentence fragment
      prior to the comma above.

      See below for an example which uses a simpler domain than "intelligent
      behavior" (which is quite difficult to establish objective criteria
      for).

      > Machines have an existence and a set of possible
      > behaviors which does not depend on the biological
      > machines that may have inspired their construction.

      Good stuff.

      > You cannot copy the brain unless you use the same
      > materials. The difference between a copy and a
      > simulation at a functional level is one of detail.

      More good stuff.

      > A simulation is a system which is isomorphic
      > "similar in pattern" as the system it is simulating.
      > The relationship between a set of variables which
      > are deemed essential or of interest happen to have
      > the same relationship in both systems.

      Very good stuff!

      > Unlike a simulated hurricane an AI program can, via
      > transducers and effectors, act upon and be acted
      > upon by the real environment.

      Yes, certainly I agree.

      Let's stay generally with "simulation" vs "non-simulation".

      For example, can you "simulate" rocks?

      You often hear statements such as "A simulated rock breaks no
      windows". Certainly statements like this implies a certain type of
      "simulation". Maybe we could call these "virtual simulation"?

      (keep this in mind as we look at another type of simulation)

      But here is an example of another *type* of simulated rock:

      Cubic Zirconia is a simulated diamond. It's rather easy to test for
      the difference between CZ and diamonds. But, we could even take it a
      step further:

      Low pressure diamond (Chemical Vapor Deposition or High Pressure, High
      Temperature if you prefer) is simulated "natural diamond".

      It's much more difficult to test for differences given examples
      produced from the defining processes which produced them.

      As it turns out, there are tests which can distinguish "natural" from
      "CVD" or "HPHT" diamonds. But, what if this were not the case? What if
      a synthetic process could produce stones indistinguishable from the
      "natural diamond" process. One for which there is *no* test possible
      for sorting examples of natural from synthetic diamonds?

      Certainly in "simulated" cases such as these, knowledge of the
      difference in the processes which resulted in the examples is the only
      thing which defines "simulated" from "non-simulated". The
      ***examples*** themselves contain no "information" as to the
      differences in these processes at all.

      Now, with respect to your original "pre-comma" statement above, I tend
      to believe in a *type* of "intelligent behavior" for which there are
      *no* test, for which the "examples" contain no information for sorting
      by "process".

      In other words, I agree with you as far as I can tell (the question
      hinges on the *type* of simulation we are discussing and not mixing
      the two).

      JJG
    • Bill Modlin
      Simulation/copy/model whatever... people seem to use these words in a variety of ways to make whatever argument they fancy, and forget that the words have
      Message 2 of 8 , Jan 1, 2008
      • 0 Attachment
        Simulation/copy/model whatever...  people seem to use these words in a variety of ways to make whatever argument they fancy, and forget that the words have pretty broad definitions.

        True a computer model of rainfall patterns doesn't get anything wet.

        But I often simulate rain with a sprinkler, and it does a pretty good job of getting the grass wet.  I might control the sprinklers with a computer for a more effective or economical simulation.

        A mathematical simulation of a rock impacting glass doesn't leave fragments to slice your hands.

        But there are simulated rocks in our neighbor's flower beds.  I think they are cast of plaster, or maybe some sort of concrete mixture.  They certainly are not real rocks.  Nevertheless they would probably break a window if hurled.  Even if thrown by a computer controlled robot simulating the actions of a human vandal.

        Model airplanes routinely fly in neighborhood parks.

        A good computer simulation of an intelligent entity (and not just of some limited aspect of such an entity) would actually think and solve real problems.

        What a simulation or copy or model can do depends on what capabilities or attributes you happen to be implementing.   Not on what you call it.  Blanket statements about "what a simulation can't do" are not philosophy, just silly argumentation about words, irrelevant to any real issues.

        Bill Modlin

      • Joe Legris
        ... to make whatever argument they fancy, and forget that the words have pretty broad definitions. ... grass wet. I might control the sprinklers with a
        Message 3 of 8 , Jan 1, 2008
        • 0 Attachment
          --- In ai-philosophy@yahoogroups.com, Bill Modlin <wdmodlin@...> wrote:
          >
          > Simulation/copy/model whatever... people seem to use these words in a variety of ways
          to make whatever argument they fancy, and forget that the words have pretty broad
          definitions.
          >
          > True a computer model of rainfall patterns doesn't get anything wet.
          >
          > But I often simulate rain with a sprinkler, and it does a pretty good job of getting the
          grass wet. I might control the sprinklers with a computer for a more effective or
          economical simulation.
          >
          > A mathematical simulation of a rock impacting glass doesn't leave fragments to slice
          your hands.
          >
          > But there are simulated rocks in our neighbor's flower beds. I think they are cast of
          plaster, or maybe some sort of concrete mixture. They certainly are not real rocks.
          Nevertheless they would probably break a window if hurled. Even if thrown by a computer
          controlled robot simulating the actions of a human vandal.
          >
          > Model airplanes routinely fly in neighborhood parks.
          >
          > A good computer simulation of an intelligent entity (and not just of some limited aspect
          of such an entity) would actually think and solve real problems.
          >
          > What a simulation or copy or model can do depends on what capabilities or attributes
          you happen to be implementing. Not on what you call it. Blanket statements about "what
          a simulation can't do" are not philosophy, just silly argumentation about words, irrelevant
          to any real issues.
          >
          > Bill Modlin
          >

          How about this blanket statement:

          No computation can have a physical effect, except through
          non-computational, implementation-dependent physical
          properties of the machine performing the computation.
          In other words, no computation can have physical properties
          that arise from its formal operations alone. In other
          words, Searle was right.

          --
          Joe
        • Eray Ozkural
          ... I don t like this explanation because it seems to hinge on a Platonist misunderstanding of what computation is! So I can t really get from your (false)
          Message 4 of 8 , Jan 1, 2008
          • 0 Attachment
            On Jan 2, 2008 4:44 AM, Joe Legris <jalegris@...> wrote:
            > --- In ai-philosophy@yahoogroups.com, Bill Modlin <wdmodlin@...> wrote:
            > >
            > > Simulation/copy/model whatever... people seem to use these words in a variety of ways
            > to make whatever argument they fancy, and forget that the words have pretty broad
            > definitions.
            > >
            > > True a computer model of rainfall patterns doesn't get anything wet.
            > >
            > > But I often simulate rain with a sprinkler, and it does a pretty good job of getting the
            > grass wet. I might control the sprinklers with a computer for a more effective or
            > economical simulation.
            > >
            > > A mathematical simulation of a rock impacting glass doesn't leave fragments to slice
            > your hands.
            > >
            > > But there are simulated rocks in our neighbor's flower beds. I think they are cast of
            > plaster, or maybe some sort of concrete mixture. They certainly are not real rocks.
            > Nevertheless they would probably break a window if hurled. Even if thrown by a computer
            > controlled robot simulating the actions of a human vandal.
            > >
            > > Model airplanes routinely fly in neighborhood parks.
            > >
            > > A good computer simulation of an intelligent entity (and not just of some limited aspect
            > of such an entity) would actually think and solve real problems.
            > >
            > > What a simulation or copy or model can do depends on what capabilities or attributes
            > you happen to be implementing. Not on what you call it. Blanket statements about "what
            > a simulation can't do" are not philosophy, just silly argumentation about words, irrelevant
            > to any real issues.
            > >
            > > Bill Modlin
            > >
            >
            > How about this blanket statement:
            >
            > No computation can have a physical effect, except through
            > non-computational, implementation-dependent physical
            > properties of the machine performing the computation.
            > In other words, no computation can have physical properties
            > that arise from its formal operations alone. In other
            > words, Searle was right.

            I don't like this "explanation" because it seems to hinge on a
            Platonist misunderstanding of what computation is! So I can't really
            get from your (false) premises to your concllusion about formal
            properties not having "inherent" physical properties. I am not even
            sure if the statement in the conclusion is sensible or coherent.

            The correct POV here is, IMHO, that computation *is* a particular kind
            of physical process, i.e. it is *identical* to a physical macro-event,
            hence it is not fair to say that it has no physical effects, since
            every particular computation clearly is physical (and hence does have
            physical effects). On the other hand, there exist no universals, hence
            it is absurd to talk about properties of universals here.

            That is to say, we need a particularist identity theory of computation
            as opposed to a Platonist theory of computation that might imply
            certain silly kinds of nonreductionism/dualism to many people.

            Best,

            --
            Eray Ozkural, PhD candidate. Comp. Sci. Dept., Bilkent University, Ankara
            http://www.cs.bilkent.edu.tr/~erayo Malfunct: http://myspace.com/malfunct
            ai-philosophy: http://groups.yahoo.com/group/ai-philosophy
          Your message has been successfully submitted and would be delivered to recipients shortly.