AI, Philosophy, Physics and Law

o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
...Forking a thread into its more appropriate discussion group...

This is intended to be an open ended discussion exploring the natures of and relationships between Ethereum, AI, Philosophy, Physics and Law.

Please keep it civil.

Comments

  • BiodomBiodom Member Posts: 693 ✭✭✭
    @o0ragman0o

    Awesome, bookmarked, otherwise...watercooler?
  • o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
    I didn't see any category for 'Psudorandom Determanism' ;)
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    Ok, I'll start: What goes up, must come down, at least in this universe, where gravity is in play.

    Seriously, though, any discussion of AI has to start with a clearly defined context. AI is all about the specific implementation, i.e. what are the exact capabilities inherent in a particular AI manifestation. It's like Data in Star Trek Next Gen: Data pre-emotion chip, or Data post-emotion chip - big difference. There's really no such thing as "generic" AI.
  • o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
    dlehenky said:

    Ok, I'll start: What goes up, must come down, at least in this universe, where gravity is in play.

    You're trolling right? You didn't define your inertial reference frame.... ;)
    dlehenky said:


    Seriously, though, any discussion of AI has to start with a clearly defined context. AI is all about the specific implementation, i.e. what are the exact capabilities inherent in a particular AI manifestation. It's like Data in Star Trek Next Gen: Data pre-emotion chip, or Data post-emotion chip - big difference. There's really no such thing as "generic" AI.

    Alas, didn't get past 3 responses before Star Trek was mentioned :(

    AI pioneer Peter Norvig, said something like (sorry I can't find the actual quote), "Asking if computers can think is like asking if a submarine can swim. It doesn't matter, can it do the job?"

    So looking at AI framed by 'can it do the job?' makes the whole field much easier to understand. Furthermore it reduced the expectations upon AI from 'be all and end all' existentialism to being a diverse emergent ecosystem of simpler specialised funcational AI's, which is of course, just a continuation of evolutionary dynamics.

    So it's unlikely that Ethereum will ever become some self aware Skynet (it's processing power is negligible), but it is also unpredictable what kind of emergent ecosystem (Etho-system?) will arise from the Ethereum/IoT simple minded machines in such an environment.
  • BiodomBiodom Member Posts: 693 ✭✭✭
    edited May 2016
    For those interested in this topic, I suggest reading two articles in the current issue of Wired. Not sure that they are online-I have subscription.
    The first one deals with a suggestion that we will be mostly not be coding the software, but "teaching" it soon (neural nets, etc).
    The second describes an incredible AlphaGo adventure and deep analysis of what happened there. Apparently, software was able to achieve something that we call "intuition"-designed a move in Go that was not pre-computed, had never happened before and by software's own analysis was only predicted to happen with a 1:10000 chance in a similar situation in a Go game played by humans alone. Yet, AlphaGo made that move.
    Human counterpart was so shocked that he left the game for few minutes to compose himself, then lost within an hour or less. This is not the end, though, as in the next game he came up with a move that shocked the computer and was subsequently called a "Touch of G-d". Computer lost that one.

    Bottom line: the human player so enhanced himself as a player after this encounter that he did not lose a single game with humans later on, apparently.
    So, my thesis would be that we will teach AI to be more human and it will teach us a thing or two as well.
    After that, we might join or at least enhance ourselves.

    There will be no separate AI, AI would be a part of us. At least, this is my hope.
    Post edited by Biodom on
  • o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
    Just watched 'Chappie'..... meh. Giving a robot 'personality' is not the same as giving it consciousness. The overly anthropomophized cutesy factor did absolutely nothing for me and the movie added nothing more to the AI genre that wasn't already covered in 1986's 'Short Circuit' ("Johny 5 is alive")... which as a kid I enjoyed.

    The idea that human consciousness is 'substrate independent' and can be reduced to a .dat file and uploaded, is a load of BS to me, I believe in a far more mystical, or at least quantum origin to consciousness. While the processes of consciousness can be modeled well, it's still just mimicry rather than actuality.

    In the practical case though, the question must be asked if those processes will out perform human processes for the same task...('go' figure). Yes, but I'd say in the similar manner that a hydraulic ram out performs biological muscle.

    Ignoring all it's other problems, I liked Ex Machina because it demonstrated well what an AI's nature actually is like and the problem of mistaking or trying to force that to be analogous to 'human consciousness'.....

  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @o0ragman0o It may come to you as no surprise, but I do not consider consciousness to be of this earth, nor that it is partitioned off into an individual, unconnected, domain called a person, at least not exclusively. To me, consciousness and intellectual brain function are two entirely different things, although in sentient beings, there is, to one degree or another, a connection between them that's often described as intuition, inspiration, a spark of genius, an epiphany, etc.. I'm thinking perhaps you've read Amit Goswami's works?
  • o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
    @dlehenky So without throwing out all sense of material influence of consciousness, I attempt to understand it in with a kind of reductive reasoning, not in the sense of reduction into parts but by stepping down through the foundational levels of what emerges into something like human sentience and waking awareness.

    For a human we have a complex brain. That brain has specialized substructures but in general is a biological neural network. Electrical and chemical tampering of the various regions of the brain has allowed scientists to be comfortable that the brain's neuronal structures appear fundamental to human consciousness and so propose the theory of "Neural Correlates of Consciousness"(NCC). Though one notes by the term 'correlate', that the scientists are in no way prepared to peg their careers on and call it 'cause'.

    At best the human brain is a specialised application of the neural network mechanics, while the brain of other animals are also specialised according to the species and presenting different (but recognisable) behaviors and states of awareness.

    So through out the neural networks we have the propagation of weighted stimulants and structured transmissions, each with their own complex dynamics of bio/electro chemistry itself presenting high orchestration that could only be seen as having an intelligent agency beneath. And so on into the cells themselves, wherein complex and autonomous molecular machinery exist in dynamic societies actuated by the ever restructuring fields down to a quantum level... Societies that build structures like microtubules, the endo-skeletons giving cells structure, along which propagate quantum automata similar to Conways Game of Life. Along these quantum automata thing like motor proteins literally 'walk' with alternating steps of their molecular feet, towing great bags of proteins to be delivered to other needy regions in the cell.

    In neurons these quantum automatons terminate delivering quantum packets of highly structured energy directly at the sites of the synapses and collapse into classical states that give rise to the electrochemical dynamics of neurotransmission. This is Penrose and Hammeroff's theory cald 'Orchestrated Objective Reduction' or Orch-OR for short.

    Objective Reduction is Penrose's interpretation of quantum mechanics whereas the 'Orchestrated' bit is Hammeroff, an anaesthesiologist, applying that to a conscious agency. So this theory, (to which I strongly yeild) fairly places the source of consciousness directly in the structured energies at the level of quantum fields rather than at some minimal complexity of a biological structure.

    So thus my serious doubts about the 'substrate independence' of consciousness. I imagine it extremely hard if not completely pointless, to simulate a structured quantum fields of sufficient complexity to give rise to the necessary emergent simulation of electro-chemical dynamics simulations and so on up the simulation chain to an entire simulated brain.

    Given that the structures upon which those simulations must be stored and run are required to be orders of magnitude greater in size, complexity and energy for the required function and storage of state, the exercise rapidly blows out of anything that we can actually engineer. It's for this reason that I have great doubts about 'The Human Brain' project that seeks to digitally simulate such a brain from a chemical level up.

    Certainly we'll learn lots about the human brain and likely even recognized responses of this digital behemoth as 'human-ish', but isn't so much more effective and wonderful to 'have a baby'? So greater a beauty to recognise the glory and intelligence in all life?


    I'm not anti-AI. Just that having attempted to program it since I was a kid has given me a much greater appreciation of the real thing, and I think that is a healthy perspective to keep.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @o0ragman0o Wow, back to school ... :)

    At this point in my life, I've stopped really caring about the "how", for the most part, other than my own geek curiosity. I have come to view life from an exclusively spiritual point of view, which simplifies everything, for me. It's quite simple, actually ... it's simply perfect, beyond what we, in this relative world, will ever comprehend, and, indeed, weren't meant to (not to say we shouldn't strive to). I know that God *is* consciousness, a purely creative (what else can consciousness do?), absolute, infinite consciousness, and that we are all cut, in equal portions, from that fabric. Of course, having said that, consciousness is only one facet of God. The Hindu tradition recognizes countless facets and name them all with great joy, as I'm sure you know.
  • BiodomBiodom Member Posts: 693 ✭✭✭
    edited May 2016
    @o0ragman0o @dlehenky

    To simplify, for me consciousness is an emergent property of our complex brain.
    By analogy, looking at the random ant movement, you cannot imagine a result of such motion being an anthill, unless you already know that.
    Whether there is some consciousness-specific effect at the quantum level, I pretty much doubt.
    At least, it would not be human-specific.
    People already know that great apes, dolphins, elephants and, surprisingly, some birds, are capable of introspection-famous mirror experiment.
    We have twice as much brain size in comparison with chimps, most likely due to the action of a single gene (microcephalin). We get roughly one additional cell division among brain cells. It is possible that once brain reaches certain complexity, a new property 'emerges'.
    Moreover, if you are homozygous for a bad mutation in microcephalin, you get diminished brain size and severe mental retardation. Read microcephalin wiki-it is fascinating
    This does not take us out of spiritual realm, but i am discussing just the scientific facts here, a subject i know something about, while i know very little in the 'spirituality' field.

    All of it does not explain the example that I gave re Go-playing computer (alphaGo) coming up with a move that cannot be explained. Was it machine's intuition? It was NOT pre-computed or programmed and was not present in computer's database.
    Post edited by Biodom on
  • o0ragman0oo0ragman0o Member, Moderator Posts: 1,291 mod
    edited May 2016
    Biodom said:

    @o0ragman0o @dlehenky

    To simplify, for me consciousness is an emergent property of our complex brain.
    By analogy, looking at the random ant movement, you cannot imagine a result of such motion being an anthill, unless you already know that.
    Whether there is some consciousness-specific effect at the quantum level, I pretty much doubt.
    At least, it would not be human-specific.

    @biodom So when I'm thinking of 'consciousness', both in the scientific and spiritual sense, I tend to think upon the cause or source for whatever that means. For science the traditional notion of consciousness is, lets say 'presenting a lucid self actuated awareness of one's body within an environment' in contrast to 'unconsciousness' as being a state in which no awareness is present at all, not even to time passing. In this sense, consciousness can be present or absent according to a emergent properties of biological states.

    In the spiritual sense, consciousness can be thought of more or less as something physically transcendent and inclusively universal to which biology acts as a transducer into a more commonly recognised awareness and agency.

    The different natures of these consciousness paradigms you can imagine to be in the different 'magesteria' of science and mysticism with the remaining questions of whether they overlap or not. My interests definitely explore the overlap.
    Biodom said:


    People already know that great apes, dolphins, elephants and, surprisingly, some birds, are capable of introspection-famous mirror experiment.
    We have twice as much brain size in comparison with chimps, most likely due to the action of a single gene (microcephalin). We get roughly one additional cell division among brain cells. It is possible that once brain reaches certain complexity, a new property 'emerges'.
    Moreover, if you are homozygous for a bad mutation in microcephalin, you get diminished brain size and severe mental retardation. Read microcephalin wiki-it is fascinating
    This does not take us out of spiritual realm, but i am discussing just the scientific facts here, a subject i know something about, while i know very little in the 'spirituality' field.

    I didn't know about additional cell division. That's interesting. I use to assist with the schooling of microcephalic kids along with other severely impaired kids including one child with an aquired brain injury (abi) so severe that he was considered by the 'black' CT scan to be medically brain-dead. Having a self education (and a Self education) in neural networks, the ABI child's rehabilitation was particularly intriguing to me. Over about 7 months he went from being completely paralyzed with no voluntary response to saying his first word and taking his first (assisted) step. As he gained some volition first over his eye movements we could then begin to get an insight to his experience and state of awareness. It was found he could still spell and do primary school level maths (he was 11). His emotional responses appeared quite normal but most of all he seemed to have had a very hightened awareness of his environment and found a good many trivial things highly amusing as attest by his frequent outbursts of laughing. So for whatever was left of this child's brain, (he'd been hit by a car and unrevived for ~30 minutes in tropical heat), there was most definitely something of a higher order than medical science could fully explain. It was a quite different and much sadder story with the microphalic kids.

    That said, there is something of a scale invariant property to neural networks that allows something like a honey bee to be just as good a navigator as a human in complex environments. And that the intelligence and maturities of consciousness of a mouse is sensibly recognisable as sentient, rather than just some Descartian automaton.

    So I have no argument against the emergent properties of neural networks and the Neural Correlate of Consciousness, though I do wonder how many thresholds of complexity there are between a mouse and a human. I wouldn't think it needs to be many at all. For instance, a singular chain link is inflexible, yet from the linking to other chain links, the property of flexibility emerges that was not in any way a property or even the sum of it's parts. Instead, emergence comes more from relationship of parts rather than their properties. So once that property emerges there really aren't any further emergent properties, until you introduce a different relationship, i.e a short chain is still a chain, just like a long chain is a chain. So a small brain is going to display the same properties of a large brain and so to a digital representation of a brain.

    So after identifying relationships rather than components as a causal mechanism of emergent properties, we can then look at relationships within other non-neural systems such as the electro-chemical orchestration within the cells themselves in order to solve the extraordinarily complex problems of cellular metabolism itself in undeniably 'intelligent' ways.

    Recognising a problem solving intelligence at even cellular and molecular level, one can only assume its properties must be emergent from a structured quantum field. And that is where the Magisteria begin to overlap (and give rise to a veritable inflaton field of new age guff :/ ).
    Biodom said:


    All of it does not explain the example that I gave re Go-playing computer (alphaGo) coming up with a move that cannot be explained. Was it machine's intuition? It was NOT pre-computed or programmed and was not present in computer's database.

    Sorry, I'd not been able to catch up on the article you mentioned until today. Personally, I don't get Go. I've tried a couple of online games and have been left completely perplexed by it. So when they comment about some particular move, I simple don't have the experience to appreciate what they're on about. Though I did watch some of the AlphaGo games live given my interest in AI.

    But now having some cursory reading of the analysis of move 37, it seem to me that AlphaGo simply happened upon an advantage based more because of the 'conditioning' of human minds rather than anything digitally mystical. It's quite conceivable that humans playing other humans are going to develop conditioned expectations and responses which 'prune' their collective ability to explore other branches in a particular search space. In this regard, a dumber lesser conditioned machine could quite easily happen across a branch that simply hadn't occurred to humans before.
    Post edited by o0ragman0o on
  • legend218legend218 Member Posts: 5
    interesting read.
  • jdetjdet Member Posts: 16
    Please be specific while talking about neural nets. Artificial neural networks are just mathematical abstractions of organic neural networks. Failing to distinguish between them might cause a lot of confusion.
Sign In or Register to comment.