[FRIAM] the arc of ai (was Re: Whew!)

[glen ☣]

At some point, wouldn't we enter David Deutsch (or Neal Stephenson) territory?  ... where the idea is that the computation in our nervous system is mappable to the computation going on around us.  If consciousness isn't compressible because it's an artifact of that mapping, then faster neurons wouldn't make us more rational in any sense.  But it may give us super powers ... like running really fast ... or deriving nutrition from eating grass or maybe seeing a wider spectrum of color, etc.  But it wouldn't necessarily change our intent or purposeful, self-references.

I suppose we could also invoke Buzsáki or the like, as well... that the larger ("beat") frequencies of collections of neurons may not change even if the individual neurons' rate does change.

And then there's the extra-neural computation, as well, including glial cells and glucose across the BBB.  The success of your Josephson Junction ANN would become a question of the power generation and distribution, regardless of the computational question of whether the high-dimensional, multi-rate (chemical) space can be reduced to electrical signals.  (I wouldn't have gone here if I weren't currently in the throes of a massive headache. [sigh]  Will robots get headaches?)

On 05/08/2017 12:31 PM, Marcus Daniels wrote:
> Seems to me there's a question of dynamic range, temporally speaking.  In classical computers, that is dealt with by separating exponents and mantissas as in floating point arithmetic.  If everything compresses by 7 order of magnitude, then perhaps it would just be a matter of adding 7 more digits (e.g. bits) of precision to the exponent.  Then it would be faster and cover the relevant part of the dynamic range of the environment.


-- 
☣ glen