RE: Threats to the Singularity.

From: Dani Eder (
Date: Fri Jun 14 2002 - 11:30:42 MDT

It seems to me that the seed AI code size will
be a function of at least the following:

1) Processor speed
2) Storage capacity
3) Learning/growth time

In general, the more powerful the computer
and the more time you have, the smaller the seed
AI can be.

For example, given a computer powerful enough to
run a simulation of 10^11 neurons (a human brain's
worth), you should be able to reduce the code to
simulate a single neuron at a time, then repeat
10^11 times in a loop. Then feed it sensory inputs
for 20 years and watch the intelligence develop.

At the other extreme, when you want the computer
power minimized, as well as the learning time,
you will have to incorporate an adult level of
knowledge into your code, so you will have Gigabytes
of knowledge incorporated from the start in the
form of clever algorthims, knowledge bases, etc.


Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:39 MDT