Tuesday, October 7, 2014

a response in a thread on kurzweilai forums

[quote]Most people imagine that shortly after the first human level AGI is created an SAI is destined to emerge shortly there after, one that is absolutely unfathomable to our puny little meat brains. When the truth is our brains could put Watson to utter shame (assuming they had a way to assimilate the data rapidly) if they were redesigned and optimized for a such narrow task..[/quote]
 While difficult, comparative genetics can show the modifications from simpler mammals like rodents, to primates to humans, and even among humans the differences bestowing greater intellect.   It remains an open question how straightforward the modifications are, is there a clear path or if they're highly specialized and totally unique  with each leap in capacity.   If it turns out that in general there has been a straightforward route of modifications to the neural wiring and computing algorithms employed in animals as their intelligence increases, then it will be possible to extrapolate the design choices forward to their theoretical limits.
[quote]If you speed up a monkey 100 fold, that monkey will just do monkey things 100x faster. I don't see why it should be all that different for humans.
[/quote]
Given 100xspeed up a human could master countless fields of science, with the plasticity of a child he could easily attain human native performance in countless languages, and with enhanced memory he could have encyclopedic knowledge of all he masters.    A monkey is a sublinguistic entity and thus limited in what it can endeavour, discover and accomplish, a human is not bound by such limitations.
[quote]Trying to turn a digital computer into a brain, though, that is an entirely different prospect. I would compare it to trying to modify a candle through step by step alterations (each stage of which functions well enough to be useful) until you have an electric light bulb. One can obviously make trivial similarities between candles and light bulbs (both provide illumination) just as one can with computers and brains (both process information) but the differences in design principles are so profound there really can be no evolution of one into the other.[/quote]
The difference is that it is assumed what the brain is performing is computation, if it is then like flight happens with planes without turning them into birds, a different architecture that is universal can also perform the same computation, because algorithms are substrate independent.    Of course if it is doing some spooky or magical thing that is not computation then it can only be approximated.
That is when you go into the camps of those who say physics is computable, including quantum physics, and not even quantum computers are more powerful than traditional turing machines albeit faster at certain tasks, and those who say physics is not computable.   Even in the case of physics not being computable, it is said that it is not possible to physically build a computer that surpasses a turing class  computer, so the brain would have to not be a computer at all or  if so it would need to be some manner of hypercomputer or such claims would have to be flawed.

[quote]The difference between the the most idiotic of our species and our most accomplished has little if anything to do with the speed with which the brain "computes". Instead it has everything to do with the integrated models that are physically instanced within each particular brain.[/quote]
Both white matter quality and grey matter volume have been linked to iq.  White matter increases speed, but within a simulation it is possible for the communication to appear to occur instantaneously for the simulated entity.   As for increased grey matter, Increasing the number of elements is also trivial, and in a few decades we may have countless exabytes allowing for countless number of elements.
There are other variations such as the size of various divisions within the brain, in sensory areas making those with less allotment more suceptible to illusions in said modality.   There are also countless variations in many parts of the system at a molecular level, making some more prone to hostility, with worse or better memory, more faithful, easier to get addicted, etc

No comments:

Post a Comment