Re: When does it pay to play (lottery)?

From: Eliezer Yudkowsky (
Date: Mon Jan 24 2005 - 17:02:01 MST

Ben Goertzel wrote:
>>If by GP you mean genetic programming, then natural selection coughed up
>>human intelligence starting with considerably less sophistication (and
>>certainly less direction) than at least some modern-day GP frameworks.
> I'm sorry, Eli, but on this point you're extremely wrong. The subtlety and
> sophistication of the human genetic and proteomic apparatus exceeds that of
> any existing genetic programming technology by very many orders of
> magnitude. Having extensively used GP to analyze human genetics data
> (, I'm very well-positioned to make judgments on this
> particular matter.

I said *starting with* and the terminology was deliberate. I was referring
to the initial replicator or initial replicating hypercycle.

> Roughly speaking you could say that: Genetic Algorithms and Genetic
> Programming are to natural evolution, as backpropagation neural nets or
> Hopfield neural nets are to the human brain. GA/GP's are not going to do
> what natural evolution did, any more than a backprop NN is going to display
> the intelligence of a human neuronal network.

Backprop NNs don't concern me because they are (usually) not open-ended
optimization processes. A GA is an open-ended optimization process. So is
one replicator or replicating hypercycle.

Unintelligence coughed up intelligence once, and the spark of that fire
(before it accumulated all the complexity to which you refer) was less
sophisticated and certainly less directed than today's GAs, which come into
existence carrying fully-formed complexity that might have taken a billion
years to evolve on Earth, or, in some cases, would not be accessible to
natural selection at all.

The computational path from unintelligence to intelligence exists. It was
climbed once in the total absence of thoughtful design, just from steady
optimization pressure. Don't tell me it can't happen again, not unless you
can calculate your answer. If you look in George Williams' classic
"Adaptation and Natural Selection" (circa 1966!) you will find more math
than you have now, and it is math indicating surprisingly small cumulative
selection pressures from the first replicator to humankind. First rule of
gun safety: A gun is always loaded.

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT