Page 1 of 3 123 LastLast
Results 1 to 30 of 71

Thread: Artificial and Organic Intelligence Questions

  1. #1
    Join Date
    Apr 2010
    Posts
    418

    Artificial and Organic Intelligence Questions

    Hi all- I have a few questions about AI after reading the Ian McEwan's new novel "Machines like me" ..... I thought someone here might help me :-)
    Lets assume no magical thinking/souls etc! The mind is the product of the brain.

    1. Is the brain a computer? Is the mind the software? Is the software/hardware a false dichotomy? It seems we model consciousness on the latest technology- a steam engine/ a calculating machine/ a telephone exchange/ a computer/ the internet etc
    2. Is the brain a digital computer? What other type is possible? Is it a combination of digital and analog? Is all analog just digital with a finer grain? Is a 'neural network' a digital computer or can it just be 'modelled' by a digital computer.
    3. Does 'chaos'/ non-linear feedback loops mean that the human mind could be deterministic yet unpredictable. i.e. free will doesn't exist but predicting human behaviour is impossible other than statistically by plotting 'strange attractors'
    4. Is the 'turing test' a 'cop-out?' Can we ever know if we have built a truly conscious mind or just a model of one that acts 'as if' it is conscious? Just as we cannot know any other minds truly exist apart from our own? We only have access to our own qualia.
    5. Is an internal 'self-awareness'/ consciousness even necessary for AI? Presumably we aren't required to simply make a facsimile of a human mind?
    6. We can't define our own organic intelligence/consciousness ... how will we know if we have made an AI?
    7. Could we 'grow' a mind starting with a massive undifferentiated blob of neural networks (a foetal brain) ... whose connections could be 'pruned' as per neural darwinism.. with 15 years of nurturing/love/education?


    https://www.theguardian.com/books/20...-mcewan-review
    "It's only a model....?" :-)
    https://www.youtube.com/watch?v=m3dZl3yfGpc

  2. #2
    Join Date
    Apr 2010
    Posts
    418
    also.. interested to hear of how these ideas have been tackled in other scifi !
    "It's only a model....?" :-)
    https://www.youtube.com/watch?v=m3dZl3yfGpc

  3. #3
    Join Date
    Aug 2006
    Posts
    3,493
    Quote Originally Posted by plant View Post
    2. Is the brain a digital computer?... Is it a combination of digital and analog? Is all analog just digital with a finer grain?
    No.

    A teacher once described the difference, thus:

    At your job, you get paid by wire as an electrical signal - either digital or analogue.

    If you choose analogue, you paycheque is sent as the voltage. A 10-volt signal means you get paid $10 per hour, while an 8 volt signal means you get $8 per hour. Problem is, the wires are old and noisy and the signal is weak.

    If you choose digital, they send your hourly rate as bits - always 0 or 5 volts - forming ASCII numbers. If anything other than 0 or 5-volt signals are received, it is immediately apparent that there is an error.


    In short:
    The accuracy of analogue is sensitively dependent on what the signal is that is received.
    The accuracy of digital is only dependent on that the signal is received.

    The brain is analogue. Effect is proportional to cause.

  4. #4
    Join Date
    Apr 2007
    Location
    Nowhere (middle)
    Posts
    37,105
    Quote Originally Posted by plant View Post
    1. Is the brain a computer? Is the mind the software? Is the software/hardware a false dichotomy? It seems we model consciousness on the latest technology- a steam engine/ a calculating machine/ a telephone exchange/ a computer/ the internet etc
    We are most likely firmware, run on an incredibly complex neural net architecture.

    Is the brain a digital computer?
    No.

    What other type is possible?
    Ask any engineer old enough to use a slide rule.

    . Does 'chaos'/ non-linear feedback loops mean that the human mind could be deterministic yet unpredictable. i.e. free will doesn't exist but predicting human behaviour is impossible other than statistically by plotting 'strange attractors'
    Nothing chaotic is deterministic.

    Define "free will" sufficiently well and I can answer the rest of the question.

    4. Is the 'turing test' a 'cop-out?'
    Absolutely. It only tests the ability of the machine to lie to a human.

    Can we ever know if we have built a truly conscious mind or just a model of one that acts 'as if' it is conscious?
    What does "truly conscious" even mean?

    5. Is an internal 'self-awareness'/ consciousness even necessary for AI? Presumably we aren't required to simply make a facsimile of a human mind?
    We are not "required" to, no, but some researchers are trying to create a conscious (whatever they define that as) machine.

    6. We can't define our own organic intelligence/consciousness ... how will we know if we have made an AI?
    We have AI now. But a conscious AI... who knows? Insufficient data.

    7. Could we 'grow' a mind starting with a massive undifferentiated blob of neural networks (a foetal brain) ... whose connections could be 'pruned' as per neural darwinism.. with 15 years of nurturing/love/education?
    What is "neural Darwinism"?
    "I'm planning to live forever. So far, that's working perfectly." Steven Wright

  5. #5
    Join Date
    Jun 2005
    Posts
    13,828
    Quote Originally Posted by Noclevername View Post
    .

    Nothing chaotic is deterministic.


    I think that depends on what you mean by chaotic. In the way I usually use it (in physics), chaotic systems means systems that may be deterministic, say a many bodies problem in classical orbital mechanics, but where you can’t make predictions because a small change in the initial conditions will lead to widely divergent behavior.



    Sent from my iPhone using Tapatalk
    As above, so below

  6. #6
    Join Date
    Apr 2007
    Location
    Nowhere (middle)
    Posts
    37,105
    Quote Originally Posted by Jens View Post
    I think that depends on what you mean by chaotic. In the way I usually use it (in physics), chaotic systems means systems that may be deterministic, say a many bodies problem in classical orbital mechanics, but where you can’t make predictions because a small change in the initial conditions will lead to widely divergent behavior.



    Sent from my iPhone using Tapatalk
    A many bodies problem assumes a stable system with no external disturbances. The same could not be said of conscious minds! We get experiences and interactions that constantly disrupt us.
    "I'm planning to live forever. So far, that's working perfectly." Steven Wright

  7. #7
    Join Date
    Oct 2009
    Location
    a long way away
    Posts
    10,732
    Quote Originally Posted by plant View Post
    1. Is the brain a computer?
    Depends how you define "computer". The original meaning of the word was a person who did computation.

    Is the mind the software? Is the software/hardware a false dichotomy? It seems we model consciousness on the latest technology- a steam engine/ a calculating machine/ a telephone exchange/ a computer/ the internet etc
    I agree it is just an analogy that doesn't work very well. I don't think a hardware/software distinction makes sense.

    2. Is the brain a digital computer? What other type is possible? Is it a combination of digital and analog? Is all analog just digital with a finer grain? Is a 'neural network' a digital computer or can it just be 'modelled' by a digital computer.
    Is not clear it is a "computer" and it is hard to classify it in terms of other computing machines. But definitely not digital. Not really analog, either.

    3. Does 'chaos'/ non-linear feedback loops mean that the human mind could be deterministic yet unpredictable. i.e. free will doesn't exist but predicting human behaviour is impossible other than statistically by plotting 'strange attractors'
    Very possibly. Feedback loops (especially between different levels of processing) are probably a fundamental part of how the mind operates. (See Godel, Escher and Bach by Hostadter)

    4. Is the 'turing test' a 'cop-out?' Can we ever know if we have built a truly conscious mind or just a model of one that acts 'as if' it is conscious? Just as we cannot know any other minds truly exist apart from our own? We only have access to our own qualia.
    I think it is clear now that the Turing test is not enough. People have suggested other tests, based on images for example. But, ultimately, if an AI behaves indistinguishably from a "real" intelligence, insists it has feelings and responds emotionally to music or poetry, there will always be people saying it is not "really" feeling those things.

    5. Is an internal 'self-awareness'/ consciousness even necessary for AI? Presumably we aren't required to simply make a facsimile of a human mind?
    6. We can't define our own organic intelligence/consciousness ... how will we know if we have made an AI?
    Good questions. Possibly unanswerable.

    7. Could we 'grow' a mind starting with a massive undifferentiated blob of neural networks (a foetal brain) ... whose connections could be 'pruned' as per neural darwinism.. with 15 years of nurturing/love/education?
    In principal, probably. But the ethical issues are horrendous.

  8. #8
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    Can we convince the AI that we're intelligent? That might be more important.

  9. #9
    Join Date
    Oct 2009
    Location
    a long way away
    Posts
    10,732
    We expect an AI to exhibit intelligence far in excess of humans. I'm not sure why this is. Presumably, it is based on the fact that we can build machine learning systems (so called AI) to solve specific problems better than humans. This is partly because they are dedicated to a single task and will use all of the allocated timed resources solving it.

    If we are able to build a "real" AI then it would probably have the same quirks as a human intelligence. Getting distracted by other things, boredom, and so on. They may be no better at problem solving than a smart human.

  10. #10
    Join Date
    Apr 2007
    Location
    Nowhere (middle)
    Posts
    37,105
    Quote Originally Posted by Strange View Post
    If we are able to build a "real" AI then it would probably have the same quirks as a human intelligence. Getting distracted by other things, boredom, and so on. They may be no better at problem solving than a smart human.
    They probably won't have human quirks, but will develop quirks of their own like nothing we've experienced. Especially at first before they self-edit.
    "I'm planning to live forever. So far, that's working perfectly." Steven Wright

  11. #11
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,714
    AI is a misnomer if it implies human likeness. Mostly it is fast processing and learns rather like a slime mould only faster. Humans have emotions and use those to take decisions. Generally we do not want AI to do that but to be rational. The various dilemmas that can be foreseen, for example self driving AI, are posed as human emotional problems or , even worse for AI , as moral issues. AI can simulate emotion by programming and that might fool people in a Turing test, but it is easy to see it is not real emotion. But here I go introducing what is a real emotion? We are only just working that out for humans, we have to work on octopuses and corvids before attempting AI emotions. If we ever find a good reason to do so.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  12. #12
    Join Date
    Aug 2006
    Posts
    3,493
    Quote Originally Posted by Noclevername View Post


    Nothing chaotic is deterministic.
    As Jen points out, this is not really true.

    Programs that utilize chaotic equations, if given the same inputs, can easily reproduce identical results i.e. deterministic.

  13. #13
    Join Date
    Oct 2005
    Posts
    26,778
    Yes, there is a thing called "deterministic chaos", which means, sensitivity to initial conditions. It applies when the exact same initial condition always leads to the exact same outcome, but any tiny uncertainty in the initial condition will lead to a large uncertainty in the outcome. The idea is that the initial conditions in any real situation can generally not be known exactly, so even a deterministic treatment, when chaotic, still results in unpredictable results, like weather. It's the way that determinism shoots itself in its own foot, because even if you believe that exact deterministic treatments are possible, they still do not accomplish the desired result when chaotic. So even if you like to hold that the world is deterministic in principle, in practice it never acts that way, so the belief is essentially useless and impossible to verify in any system that is nonlinear enough to be chaotic. Note this also does not require that the system be complex-- one can encounter deterministic chaos even in quite simple systems of equations. The conclusion is that determinism is extremely limited as a relevant philosophy, yet it remains a rather widespread one.
    Last edited by Ken G; 2019-Oct-07 at 02:33 AM.

  14. #14
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    When we abandon our physical bodies and have our minds uploaded to computers, our behaviors will be chaotic but predictable. People could run tests on a copy of my mind to find out exactly how I'd respond to certain inputs. When they find a response that they like they can use those inputs to manipulate the real me.

  15. #15
    Join Date
    Apr 2007
    Location
    Nowhere (middle)
    Posts
    37,105
    Quote Originally Posted by Chuck View Post
    When we abandon our physical bodies and have our minds uploaded to computers,
    Would that count as artificial or organic intelligence?
    "I'm planning to live forever. So far, that's working perfectly." Steven Wright

  16. #16
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    A computer simulation of organic intelligence.

  17. #17
    Join Date
    Oct 2005
    Posts
    26,778
    Quote Originally Posted by Chuck View Post
    When we abandon our physical bodies and have our minds uploaded to computers, our behaviors will be chaotic but predictable. People could run tests on a copy of my mind to find out exactly how I'd respond to certain inputs. When they find a response that they like they can use those inputs to manipulate the real me.
    Not if Penrose is right and quantum states play a role in how our minds function. If that is true, then you would have to upload your mind to a quantum computer to preserve its function, but quantum mechanics has a "no cloning" theorem, which means you can never copy a quantum state-- each quantum state is an individual entity that can never be either destroyed or copied. So that means there could never also be a 'real you' other than the computerized mind you were transferred into. (Of course, the same mathematics that gives us these theorems gives us the many-worlds interpretation, so if quantum states are important for minds, then what "your mind" is is something vastly larger and more complicated than we already imagine it is.)

    And even if a classical computer suffices to mimic your mind function, a classical computer will likely obey classical determinism, which means it can exhibit deterministic chaos, which means you could never use a perfect classical copy of yourself to predict what you will do, beyond a few e-folding times of the appropriate Lyapunov timescale. Perhaps that timescale is about how long you have perfect short-term memory, i.e., how long can you hold an image in your head without losing parts of it. I'm guessing that timescale is no more than a few seconds, so I expect a perfect classical copy of a human brain would only be a reliable predictor of the original for somewhat less than a minute, at best. Unless it's something rather obvious that anyone could predict you would do!
    Last edited by Ken G; 2019-Oct-09 at 08:41 PM.

  18. #18
    Join Date
    Jun 2003
    Posts
    8,630
    Quote Originally Posted by Strange View Post
    Is not clear it is a "computer" and it is hard to classify it in terms of other computing machines. But definitely not digital. Not really analog, either.
    I'm reading this objection a lot in discussions of this kind, and it puzzles me a little.
    If the brain is not a computer, what else is it?
    It certainly isn't a digital computer, but there are other options in computation. The brain/body system as a whole has many analog components, such as the endocrine system and muscle memory. Even the way neurons work and memory is stored is far from being digital in nature. And there may be some quantum element to the processing as well, although I'm not convinced Penrose has got the details right. But all these processes are a kind of computation. What is it about the mind that suggests that it is not only performing computation?

  19. #19
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    Quote Originally Posted by Ken G View Post
    Not if Penrose is right and quantum states play a role in how our minds function. If that is true, then you would have to upload your mind to a quantum computer to preserve its function, but quantum mechanics has a "no cloning" theorem, which means you can never copy a quantum state-- each quantum state is an individual entity that can never be either destroyed or copied. So that means there could never also be a 'real you' other than the computerized mind you were transferred into. (Of course, the same mathematics that gives us these theorems gives us the many-worlds interpretation, so if quantum states are important for minds, then what "your mind" is is something vastly larger and more complicated than we already imagine it is.)

    And even if a classical computer suffices to mimic your mind function, a classical computer will likely obey classical determinism, which means it can exhibit deterministic chaos, which means you could never use a perfect classical copy of yourself to predict what you will do, beyond a few e-folding times of the appropriate Lyapunov timescale. Perhaps that timescale is about how long you have perfect short-term memory, i.e., how long can you hold an image in your head without losing parts of it. I'm guessing that timescale is no more than a few seconds, so I expect a perfect classical copy of a human brain would only be a reliable predictor of the original for somewhat less than a minute, at best. Unless it's something rather obvious that anyone could predict you would do!
    I'm pretty sure that if you give two functionally identical classical computers the same program and the same data you'll get the same output no matter how chaotic it might be or how long it runs.

  20. #20
    Join Date
    Jun 2005
    Posts
    13,828
    Quote Originally Posted by Chuck View Post
    I'm pretty sure that if you give two functionally identical classical computers the same program and the same data you'll get the same output no matter how chaotic it might be or how long it runs.
    I agree that's true, but I fail to see how that relates to what Ken G was saying. He was talking about whether a computer could copy the mind, not whether two computers could function the same way, as far as I can tell. Is there something I'm missing here?

    ETA: There is also the issue that quantum effects could lead to slightly different results for identical classical computers, so I means "ignoring quantum effects."
    As above, so below

  21. #21
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    If my mind were uploaded to a classical computer then that me could be simulated exactly by another such computer.

  22. #22
    Join Date
    Apr 2011
    Location
    Norfolk UK and some of me is in Northern France
    Posts
    8,714
    Quote Originally Posted by eburacum45 View Post
    I'm reading this objection a lot in discussions of this kind, and it puzzles me a little.
    If the brain is not a computer, what else is it?
    It certainly isn't a digital computer, but there are other options in computation. The brain/body system as a whole has many analog components, such as the endocrine system and muscle memory. Even the way neurons work and memory is stored is far from being digital in nature. And there may be some quantum element to the processing as well, although I'm not convinced Penrose has got the details right. But all these processes are a kind of computation. What is it about the mind that suggests that it is not only performing computation?
    So the word computer expands to cover a wider range than arithmetical or binary engines. The brain only works in a body, because of the chemical messages and links to the gut and nervous system. The idea of probability effects or chaotic states plus the ideas of epigenetics mean this computer is vastly different from any Turing type machine. So we use the word loosely in different contexts. Today we can escape the idea that the brain is everything and can be mapped. It is at least a whole person map now.
    sicut vis videre esto
    When we realize that patterns don't exist in the universe, they are a template that we hold to the universe to make sense of it, it all makes a lot more sense.
    Originally Posted by Ken G

  23. #23
    Join Date
    Jun 2005
    Posts
    13,828
    Quote Originally Posted by Chuck View Post
    If my mind were uploaded to a classical computer then that me could be simulated exactly by another such computer.
    Sure, but that ‘me’ would not be an exact simulation you, so the second computer would only be a perfect copy of an imperfect copy.


    Sent from my iPhone using Tapatalk
    As above, so below

  24. #24
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    If I abandoned my physical body and had myself uploaded, that computer version of me would be the only me. It would not be identical to the mind I left behind, but then tomorrow's mind in a biological body would not have been identical to my current mind anyway.

  25. #25
    Join Date
    Jun 2003
    Posts
    8,630
    You are never identical to past versions of yourself, so the concept of identity is a difficult one in this context.

    But you would not want to be copied in an inaccurate fashion, so that (for instance) your favourite colour changes from green to blue.

  26. #26
    Join Date
    Oct 2005
    Posts
    26,778
    Quote Originally Posted by Chuck View Post
    If I abandoned my physical body and had myself uploaded, that computer version of me would be the only me. It would not be identical to the mind I left behind, but then tomorrow's mind in a biological body would not have been identical to my current mind anyway.
    But you said one could use such a simulated version to predict what you would do, that's what I was saying would only work to a limited degree because computers are expressly designed to operate on exact input data and minds are not.
    I'm pretty sure that if you give two functionally identical classical computers the same program and the same data you'll get the same output no matter how chaotic it might be or how long it runs.
    To whatever extent that is true, then if Penrose is right about the mind, no computer of the nature you describe could ever function like a mind. I suspect that is also true even if Penrose is wrong and quantum mechanics plays no role in mind function, because of the way minds use "fuzzy logic" rather than digital logic. To put your mind into a computer, I suspect the first job will be to make computers that do not have the property you just mentioned.
    Last edited by Ken G; 2019-Oct-10 at 06:06 PM.

  27. #27
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    If the mind uses fuzzy logic then the classical computer software would have to simulate fuzzy logic closely enough for practical purposes even though it can't do true fuzzy logic.

  28. #28
    Join Date
    Oct 2005
    Posts
    26,778
    Quote Originally Posted by eburacum45 View Post
    You are never identical to past versions of yourself, so the concept of identity is a difficult one in this context.
    Yes, a lot of philosophers have thought about the concept of identity as it applies to inanimate objects (like the famous "Theseus' ship" question), but I have not seen nearly as much work done on the identity of the mind. This is quite ironic, because we generally hold that only a mind has an "identity" associated with it in the first place, so we should not bother at all with the identity of a ship! But one could imagine a similar question being applied to the mind-- if I were bombarded with radiation that changed the DNA of all my cells but did not immediately kill me, and if it transfigured me so I looked not at all like I used to, and if the trauma erased my memory and completely changed my personality, to what extent am I "still me", just changed, and to what extent am I now a completely different person with no connection to the previous one? On the topic of artificially recreating my identity, what would need to be the same for it to count as me and not someone new?

    Of course as many people know, for me "identity" is our word, our concept, so the only answer to the above is "whatever we decide." But many other people seem to imagine that "identity" is an actual thing that is not simply up to us to define however makes sense to us.
    Last edited by Ken G; 2019-Oct-10 at 08:03 PM.

  29. #29
    Join Date
    Oct 2005
    Posts
    26,778
    Quote Originally Posted by Chuck View Post
    If the mind uses fuzzy logic then the classical computer software would have to simulate fuzzy logic closely enough for practical purposes even though it can't do true fuzzy logic.
    But that's just it, simulating it for practical purposes may require taking indistinguishable input and arriving at distinguishable outcomes, as do chaotic systems and, it seems likely to me, minds. Whether or not that is a crucial aspect of a system as complex as a mind is unclear, but I suspect it is.

  30. #30
    Join Date
    Jan 2002
    Location
    The Valley of the Sun
    Posts
    9,535
    There might be lots of reasons that it won't work and we might not even know that it's not working. The uploads might have some kind of machine consciousness and not remember anything different from their previous lives. Or it might work just fine.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •