Coincidental post I wrote yesterday:
It may not be possible to imitate a human mind computationally, because awareness may be driven by aesthetic qualities rather than mathematical logic alone. The problem, which I call the Presentation Problem, is what several outstanding issues in science and philosophy have in common, namely the Explanatory Gap, the Hard Problem, the Symbol Grounding problem, the Binding problem, and the symmetries of mind-body dualism. Underlying all of these is the map-territory distinction; the need to recognize the difference between presentation and representation. Because human minds are unusual phenomena in that they are presentations which specialize in representation, they have a blind spot when it comes to examining themselves. The mind is blind to the non-representational. It does not see that it feels, and does not know how it sees. Since its thinking is engineered to strip out most direct sensory presentation in favor of abstract sense-making representations, it fails to grasp the role of presence and aesthetics in what it does. It tends toward overconfidence in the theoretical.The mind takes worldly realism for granted on one hand, but conflates it with its own experiences as a logic processor on the other. It’s a case of the fallacy of the instrument, where the mind’s hammer of symbolism sees symbolic nails everywhere it looks. Through this intellectual filter, the notion of disembodied algorithms which somehow generate subjective experiences and objective bodies, (even though experiences or bodies would serve no plausible function for purely mathematical entities) becomes an almost unavoidably seductive solution. So appealing is this quantitative underpinning for the Western mind’s cosmology, that many people (especially Strong AI enthusiasts) find it easy to ignore that the character of mathematics and computation reflect precisely the opposite qualities from those which characterize consciousness. To act like a machine, robot, or automaton, is not merely an alternative personal lifestyle, it is the common style of all unpersons and all that is evacuated of feeling. Mathematics is inherently amoral, unreal, and intractably self-interested – a windowless universality of representation. A computer has no aesthetic preference. It makes no difference to a program whether its output is displayed on a monitor with millions of colors, or buzzing out of speaker, or streaming as electronic pulses over a wire. This is the primary utility of computation. This is why digital is not locked into physical constraints of location. Since programs don’t deal with aesthetics, we can only use the program to format values in such a way that corresponds with the expectations of our sense organs. That format of course, is alien and arbitrary to the program. It is semantically ungrounded data, fictional variables. Something like the Mandelbrot set may look profoundly appealing to us when it is presented optically as plotted as colorful graphics, but the same data set has no interesting qualities when played as audio tones. The program generating the data has no desire to see it realized in one form or another, no curiosity to see it as pixels or voxels. The program is absolutely content with a purely quantitative functionality – with algorithms that correspond to nothing except themselves. In order for the generic values of a program to be interpreted experientially, they must first be re-enacted through controllable physical functions. It must be perfectly clear that this re-enactment is not a ‘translation’ or a ‘porting’ of data to a machine, rather it is more like a theatrical adaptation from a script. The program works because the physical mechanisms have been carefully selected and manufactured to match the specifications of the program. The program itself is utterly impotent as far as manifesting itself in any physical or experiential way. The program is a menu, not a meal. Physics provides the restaurant and food, subjectivity provides the patrons, chef, and hunger. It is the physical interactions which are interpreted by the user of the machine, and it is the user alone who cares what it looks like, sounds like, tastes like etc. An algorithm can comment on what is defined as being liked, but it cannot like anything itself, nor can it understand what anything is like. If I’m right, all natural phenomena have a public-facing mechanistic range and a private-facing animistic range. An algorithm bridges the gap between public-facing, space-time extended mechanisms, but it has no access to the private-facing aesthetic experiences which vary from subject to subject. By definition, an algorithm represents a process generically, but how that process is interpreted is inherently proprietary. Thanks, Craig On Friday, August 16, 2013 3:21:11 PM UTC-4, cdemorsella wrote: > > Telmo ~ I agree, all the Turing test does is indicate that a computer, > operating independently -- that is without a human operator supplying any > answers during the course of the test -- can fool a human (on average) that > they are dialoging with another person and not with a computer. While this > is an important milestone in AI research -- it is just a stand in for any > actual potential real intelligence or awareness. > > Increasingly computers are not programmed in the sense of being provided > with a deterministic instruction set - no matter how complex and deep. > Increasingly computer code is being put through its own Darwinian process > using techniques such as genetic algorithms, automata etc. Computers are in > the process of being turned into self learning code generation engines that > increasingly are able to write their own operational code. > > An AI entity would probably be able to easily pass the Turing test - not > that hard of a challenge after all for an entity with almost immediate > access to a huge cultural memory it can contain. However it may not care > that much to try. > > Another study -- I think by Stanford researchers, but I don't have the > link handy though -- has found that the world's top super computers > (several of which they were able to test) are currently scoring around the > same as an average human four year old. The scores were very uneven across > various areas of intelligence that the standardized IQ tests or four year > olds tries to measure, as would be expected (after all a super computer is > not a four year old person). > > Personally I think that AI will let us know when it has arisen by whatever > means it chooses to let us know. That it will know itself what it wants to > do, and that this knowing for itself and acting for itself will be the > hallmark event that AI has arrived on the scene. > > Cheers, > -Chris D > . > > *From:* Telmo Menezes <[email protected] <javascript:>> > *To:* [email protected] <javascript:> > *Sent:* Friday, August 16, 2013 8:04 AM > *Subject:* Re: When will a computer pass the Turing Test? > > On Fri, Aug 16, 2013 at 3:42 PM, John Clark <[email protected]<javascript:>> > wrote: > > On Wed, Aug 14, 2013 at 7:09 PM, Chris de Morsella > > <[email protected]<javascript:> > > > > wrote: > > > >> > When will a computer pass the Turing Test? Are we getting close? Here > is > >> > what the CEO of Google says: “Many people in AI believe that we’re > close to > >> > [a computer passing the Turing Test] within the next five years,” > said Eric > >> > Schmidt, Executive Chairman, Google, speaking at The Aspen Institute > on July > >> > 16, 2013. > > > > It could be. Five years ago I would have said we were a very long way > from > > any computer passing the Turing Test, but then I saw Watson and its > > incredible performance on Jeopardy. And once a true AI comes into > existence > > it will turn ALL scholarly predictions about what the future will be like > > into pure nonsense, except for the prediction that we can't make > predictions > > that are worth a damn after that point. > > I don't really find the Turing Test that meaningful, to be honest. My > main problem with it is that it is a test on our ability to build a > machine that deceives humans into believing it is another human. This > will always be a digital Frankenstein because it will not be the > outcome of the same evolutionary context that we are. So it will have > to pretend to care about things that it is not reasonable for it to > care. > > I find it a much more worthwhile endeavour to create a machine that > can understand what we mean like a human does, without the need to > convince us that it has human emotions and so on. This machine would > actually be _more_ useful and _more_ interesting by virtue of not > passing the Turing test. > > Telmo. > > > John K Clark > > > > -- > > You received this message because you are subscribed to the Google Groups > > "Everything List" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to [email protected]. <javascript:> > > To post to this group, send email to > > [email protected].<javascript:> > > Visit this group at http://groups.google.com/group/everything-list. > > For more options, visit https://groups.google.com/groups/opt_out. > > -- > You received this message because you are subscribed to the Google Groups > "Everything List" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. <javascript:> > To post to this group, send email to [email protected].<javascript:> > Visit this group at http://groups.google.com/group/everything-list. > For more options, visit https://groups.google.com/groups/opt_out. > > -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out.
