From: Eugen Leitl (eugen@leitl.org)
Date: Thu Jun 13 2002 - 12:00:39 MDT
On Thu, 13 Jun 2002, Mike & Donna Deering wrote:
> I'm not an expert on this, so I rely on others numbers. Kurzweil
Oh, that great Kurzweil expert again. That guy is just so amazing.
> estimates that the amount of DNA information specifying the brain
> design is roughly equivalent to MS WORD.
Here's your microgram of dry DNA coding for the CNS. Now make me a baby.
(You don't have the rest of it nor the wet context to go with it? Well, we
don't have the hardware container for your AI seed bits, either. You do
have these AI bits, right?).
What is 2^(length of MS Word in bits) again? How much of that space codes
for viable AIs? What's your chances of hitting that randomly (because
coding is so tight there's no gradient in sequence space to guide you)?
Depending on how supportive the context is the seed can be small. But the
smaller it is, the harder is it to devise. Darwinian evolution solved this
by starting with simple working systems, and optimized stochastically for
billenia on population sizes which range from impressive to astonishing.
You can't use any of this, because your context is completely different,
so that sequence of bits makes zero sense there.
If these are computation equivalents, how many ops are that? Granted not
all of them are relevant, but can you tell which are which? Perhaps we
should ask Kurzweil.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:39 MDT