Re: Sentient vs non-sentient super general AI

From: Eliezer Yudkowsky (
Date: Wed Jun 02 2004 - 07:46:19 MDT

Ben Goertzel wrote:
> Philip Sutton wrote:
>>It appears that you now are wanting to create a non-sentient super,
>>general AI (optimisation process) rather than a sentient
>>super, general AI (optimisation process).
> I'm not sure that his proposed optimization process has to be all that
> *general*. Perhaps it has to be general in its potential applicability,
> but it seems to me that, in accordance with his proposal, it has to be
> applied only to some specific problems, viz
> -- inferring the "collective volition" of humanity
> -- estimating which actions are most likely to lead to the fulfillment
> of this collective volition

This obviously requires general intelligence, and a huge amount of power
that can only come from recursive self-improvement. But I suppose that is
what you meant by general applicability but not general application. There
is really no such thing as a "general application"; every optimization
process is doing *something*.

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

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:47 MDT