[ExI] The symbol grounding problem in strong AI

[Stefano Vaj]

2010/1/10 Stathis Papaioannou <stathisp at gmail.com>:
> No-one claims that the brain is a digital computer, but it can be
> simulated by a digital computer. The ideal analogue computer cannot be
> emulated by a digital computer because it can use actual real numbers.
> However, the real world appears to be quantised rather than
> continuous, so actual analogue computers do not use real numbers. And
> even if the world turned out to be continuous factors such as thermal
> noise would make all the decimal places after the first few in any
> parameter irrelevant, so there would be no need to use infinite
> precision arithmetic to simulate an analogue device.

I think there is little doubt that organic brains do compute things -
besides exhibiting other features, such as burning glucides, which I
take not to be of the essence in what we seek in a brain emulation.
The fact that we can deal with ordinary arithmetics is a good enough
example, I think.

They do it rather poorly in some areas, and are much better in other.
As to the digital/analog divide, I believe it has been shown well
enough by Wolfram etc. that analog computers cannot do anything
special that digital computers could not do. As to the quantum
computing angle, organic brains are no better than digital computers
in resolving classical problems which quantum computing should
resolve, so supposing that they profit from quantum effects is
tantamount  to supposing that they profit from dark energy.

What else remains to be said? Once the cerebral computations have been
emulated, the issue of whether the emulation is "conscious" is not
really different from wondering if it has a soul. The answer is social
and cultural, not "factual".

-- 
Stefano Vaj