Computation as Shadow of Consciousness
I’m borrowing a (fantastic) image from
Incredible Shadow Art Created From Junk by Tim Noble & Sue Webster.
to make a point about the Map-Territory distinction and how it pertains to simulated intelligence.
Although a computer simulation produces an output over time, that does not mean that it is four dimensional (Time can be an additional dimension on any number of dimensions – a cartoon can be 2D but still last fifteen minutes.) By using the terms N and N-1 dimensional, I’m trying to make the point that no matter how extensive the measurements we make and how compelling their coherence seems to be, it can still be one dimension flatter than the original without our knowing it.
In the case of consciousness, I would say that it absolutely cannot be defined or described quantitatively, so that it truly is N dimensional, or trans-dimensional. Whatever number of dimensions we want to ascribe to some aspect of consciousness, that description will always be n-1 to the real thing. That is, in my understanding, the nature of representation – a destructive reduction of a presentation through a lower level (n-1 or n-x) medium which can be reconstructed by at the higher level if, and only if, there is a higher level interpreter.
Some might object to the metaphor on the grounds that computation is not like a shadow, since changing a computation has predictable effects, while changing a shadow’s appearance does not effect the object. That’s true, but again, in this case, I am using N dimensional phenomena, so that interactiveness is part of the conservation of the temporal, before and after/cause and effect axis. In such a scenario, the flatland effect itself is modified somewhat and there are more dimensions shared. More shared dimensions = more conservation of sensory-motive agreement, however, there is also more that is not shared (feeling, sensations, colors, understanding, for example).
Some who are more familiar with the spectacular capabilities of cutting edge computation might object to the over-simplification, and say that I am criticizing an older generation of approaches to machine learning. There is some truth to that, but in the rarefied air of higher math, we get into what I call the super-impersonal level of intelligence. By comparison, the super-personal level of awareness could be described as mythic or poetic. Synchronicity can be concentrated through divination techniques like Ouija boards and Tarot cards. The Platonic ‘realm’ accessed through ultra-sophisticated computation are, in my view, a dual of that kind of divination, except rooted in the generic and repeatable rather than the instantaneous and volitional. The computer is an oracle to impersonal truths about truth itself, and can build locally applicable strategies from there, however, it cannot factor in any kind of personal feeling or intention.
The computational oracle is much more seductive than divination, since it does not leave the interpretation up to the audience. The computational oracle’s output is to be interpreted as objective fact…which is tremendously useful of course, when we are talking about objects. The danger is when we start believing what it has to say about subejcts. As correct as the computer is about objective truth, it is equally incorrect about people. In the same/opposite way, the Ouija board is neither correct nor incorrect, but offers possibilities that tantalize the imagination.
Recent Comments