Lúcio de Souza Coelho wrote:
On 10/6/06, Matt Mahoney <[EMAIL PROTECTED]> wrote:
(...)
The "program" can't be larger than the DNA which describes it. This
is at
most 6 x 10^9 bits, but probably much smaller because of all the noise
(random mutations that have no effect), redundancy (multiple copies of
genes), noncoding DNA (about 98%), and the fact that most of the DNA
codes
for things other than the brain. The actual complexity of the "program"
(not including learned information) is probably closer to 10^6 bits.
(...)
Saying that all it is needed for building a brain is in the DNA is an
understatement, I am afraid. Even at biomolecular level, DNA is not
the only thing controlling ontogenesis or even cell processes; there
is a lot a proteins involved in that too, some of them even *changing*
the DNA. (Turing Machines anyone?) Also, the ontogenic process is
heavily dependent on the environment, from purely local chemical
conditions to large-scale physical fields (for instance, gravity
itself is known to influence ontogenesis). Therefore, arguably part of
the information is stored not in the DNA, but in the environment
around it. At the present moment we have no idea about how much that
"environmental storage" influences the process.
It's an underestimate, but I'd be surprised if it were off by as much as
a factor of 2.
OTOH, saying that "this is the amount of information needed to structure
the brain" doesn't say anything about how much work doing the
transformation from stored form to implemented form is going to be. To
allow for inefficiencies here we should probably estimate that the
"estimate" is off by a factor of 10. This allows for a form of storage
to be used that is less efficient, but easier to extract. This, then,
yields an estimate of 6*10^10 as a more reasonable target.
However, the more interesting question is "how much information is
required to *implement* a brain with consciousness and semantic areas
that cannot be distinguished from those of a human". (N.B.: An
experimental proof of this should NOT be attempted. You will end up
with a psychotic due to limited sensory inputs.) Again, being VERY
generous, I'll go up by a power of 10 and say 6*10^11...and then
backtrack. You don't need those sections of the brain that keep the
physical sensations working, etc. Divide by 2 to get 3*10^11.
N.B.: This is not what would be desired for an AI! This is merely a
model to help one in thinking about what the computational requirement
for such might be.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/[EMAIL PROTECTED]
