From: Darin Sunley (rsunley@escape.ca)
Date: Wed Jul 07 1999 - 22:19:55 MDT
"Eliezer S. Yudkowsky" wrote:
> Billy Brown wrote:
> >At the opposite end of the spectrum you arrive at
> > software that elegantly models the abstract processes of the mind,
[snip]
> (maybe 10^13 LOC?).
>
> Disagreement - a truly elegant description of the algorithms, perhaps as
> instructions to genetic-algorithm "compilers" or self-modifying code,
> shouldn't be using any more information than there is in our DNA: 750M, tops.
One problem. Our DNA code only describes a human infant given a human adult.
A) The genetic code doesn't describe the human adult required to "compile" the
DNA.
B) A human infant isn't even a human adult. As I understand it, a human infant IS
hardwired to learn LOTS of environmental stuff very quickly, but you need the
description of the whole environment, as well, as (presumably) human quality
agents already in the environment.
This is the major problem, that the development of a mind from infant to human
quality requires existing human quality minds.
Piaget established reasonably firmly that human infants, at various stages up to
about 7-8 years, are not merely less knowledgeable adults, but their brains (basic
operations like perception, causal analysis) work in completely different ways
then adults. To describe an adult brain with adult knowledge/perceptual structures
probably requires a lot more data then to describe an infant's mind, which (as I
understand it) is array upon array of undifferentiated "learning" structures.
Darin Sunley
rsunley@escape.ca
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:04:25 MST