Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

  1.  
    Posted By: pcstru
    Posted By: Andrew PalfreymanI think we're basically agreeing. Your model is
    language->intelligence
    and my model is
    symbols->language->intelligence

    An example of a symbol is a word, but there are many others.
    A concept is another type of symbol, even if it has no name. It just needs to be internally identifiable.

    As for epigenetics, nobody has fully cracked the code yet. But clearly it's real.


    I'm not sure we are. For a start you have now moved to "intelligence" and I was specifically talking about "human intelligence".

    I like the definition of intelligence as the ability to solve a given problem in a given time. On that basis we have animal intelligence - they solve the problem of obtaining food, or building nests or ... well you get the idea. Do they need symbols for that? Language?

    Specifically human intelligence (compared to (say) animal intelligence) is predicated on language; but intelligence itself is just problem solving and animals seem to do that without language so intelligence itself is not predicated on language - just human intelligence.
    I was implicitly indicating human intelligence.

    But you make a good point. I will have to mull that over. We all know that our brains are a neocortex wrapped around a primitive mammalian brain wrapped round a lizard brain, sort of.
    •  
      CommentAuthorpcstru
    • CommentTimeNov 7th 2019
     
    Posted By: AngusBy that definition the process of evolution is intelligent, and that process is not a living thing.


    Well OK, 'evolution' might be said to be intelligent in a narrow process way if we agree that it solves problems. What problem exactly do you think it is solving?
    •  
      CommentAuthorAngus
    • CommentTimeNov 7th 2019
     
    The continuation of the process of evolution. In other words, its own survival even though it is not a living thing.
    •  
      CommentAuthorpcstru
    • CommentTimeNov 7th 2019
     
    Posted By: AngusThe continuation of the process of evolution. In other words, its own survival even though it is not a living thing.

    Evolution would not exist without human intelligence. In that context it's an idea not the living things that constitute the entirety of the examples of it's continuation. The idea of evolution solves a problem but it is not itself solving problems.
    •  
      CommentAuthorAngus
    • CommentTimeNov 7th 2019
     
    I see what you are getting at but I disagree. Evolution seems to lead toward increasing organisation and therefore is a process running counter to the direction of entropy. Therefore it exists independent of being perceived by anybody. It is a thing just as much as is gravity.

    To cast the process of decreasing entropy as the "solution to a problem" is of course to put it in human terms, as a metaphor. But that is true for any process expressed in those terms, including problem solving by animals.
    •  
      CommentAuthorpcstru
    • CommentTimeNov 7th 2019
     
    Posted By: AngusI see what you are getting at but I disagree. Evolution seems to lead toward increasing organisation

    As measured by what?

    and therefore is a process running counter to the direction of entropy. Therefore it exists independent of being perceived by anybody. It is a thing just as much as is gravity.

    No, that is life itself, not the process of evolution which is an idea.

    To cast the process of decreasing entropy as the "solution to a problem" is of course to put it in human terms, as a metaphor. But that is true for any process expressed in those terms, including problem solving by animals.

    Yes but animals eat and are 'alive'. If they did not solve the problem of food, as an individual life form they would die. The problem solving is not a metaphor from their context. And if we die, animals might persist as long as they solve the problem of obtaining food. Evolution as an idea will need to be rediscovered if human intelligence is wiped out.
    •  
      CommentAuthorAngus
    • CommentTimeNov 7th 2019 edited
     
    Posted By: pcstruEvolution as an idea will need to be rediscovered if human intelligence is wiped out.


    This is a semantic argument. I am saying that the +entity+ that we call "evolution" exists just as do any other entities that we give names to, whether they be objects or processes. The things that exist only because we give names to them are classifications of things that do not involve physical changes or processes, such as the difference between Tuesday and Wednesday.
    • CommentAuthorloreman
    • CommentTimeNov 7th 2019
     
    But what is the definition of “problem”?
    •  
      CommentAuthorAngus
    • CommentTimeNov 7th 2019
     
    Part of the anthropomorphism.
    Otherhow I'd say homeostasis and self-organisation that are intelligence-free problems
    •  
      CommentAuthorpcstru
    • CommentTimeNov 8th 2019
     
    Posted By: Angus
    Posted By: pcstruEvolution as an idea will need to be rediscovered if human intelligence is wiped out.


    This is a semantic argument. I am saying that the +entity+ that we call "evolution" exists just as do any other entities that we give names to, whether they be objects or processes.

    There is no entity. It is not semantics to point out that unicorns have no physical existence even though we have a word for them and that word has a common understanding. You can use the word evolution to refer (as Loreman did) to any systems moving from one state to another - evolution is an abstract idea. The reality of evolution is random changes to genes (the active part) and then the selection of that variation by it's suitability to the context (passive, rolling down a hill). To me it is not really intelligent problem solving because it happens by itself as a natural process. A rock rolling downhill isn't solving the problem of getting to the bottom; it rolls ... because physics.


    The things that existonlybecause we give names to them are classifications of things that do not involve physical changes or processes, such as the difference between Tuesday and Wednesday.


    I'd agree that things with no physical existence are real but it is still true that they do not have physical existence. It is not semantics to point out the difference between the idea of you giving me a thousand pounds and actually giving me a thousand of your pounds. If you don't believe me, a mere thousand pounds will be sufficient to prove the point.
  2.  
    There was a time, a time ago, when I had a dream that recurred once every year or so for a total of probably three or four times. I could not tell you much about its content except for the strong impression it left with me upon waking. That lingering feeling was powerful but utterly nameless. A whisper of its invocation came to me whenever I looked at a particular painting by Rousseau, a cartoonish representation of a tiger crouched behind a hedge in a garden with a flower.

    So, I could not begin to provide a symbol for it. Something numinous at least. Very real to me despite being utterly unidentifiable.
    •  
      CommentAuthorgoatcheez
    • CommentTimeNov 8th 2019 edited
     
    •  
      CommentAuthoralsetalokin
    • CommentTimeNov 8th 2019 edited
     
    Hidden Markov Model?
    •  
      CommentAuthorAngus
    • CommentTimeNov 8th 2019 edited
     
    That sort of thing has been around for thirty years. What is special about GPT-2? Maybe it wrote its own article?
  3.  
    Maybe it's the sheer unbridled length of the word strings it emits without any connection or input to or from any kind of connection to or input from any kind of input or connection to or from input or output or input or output except itself.
  4.  
    You want outward-looking? Then here's a neural network that teaches itself the laws of physics.
    https://www.nature.com/articles/d41586-019-03332-7

    The ability to reverse-engineer its decision processes by giving it a synthetic corpus callosum was a neat idea.
    •  
      CommentAuthoraber0der
    • CommentTimeNov 8th 2019 edited
     
    I think the dangerus part of our AI experiments is that we let them AIs learn skills but don't give them the opportunity to find their context. Those constructs are extrasocial by design.
  5.  
    profundiditty
    •  
      CommentAuthorAngus
    • CommentTimeNov 9th 2019
     
    •  
      CommentAuthorgoatcheez
    • CommentTimeNov 15th 2019