Re: AI timeframes

From: Philip Sutton (
Date: Thu Apr 15 2004 - 10:47:00 MDT

Over the last week people have been discussing some of the drivers
for take-off of advanced general artificial intelligence. This is an area
that would benefit from the application of some AGI mental power as
complex systems modelling is a key area where human minds do not
perform that well.

On Friday 9 April James/Andrew Rogers proposed a scenario where
commercial competition could lead to an AGI development 'arms race'
- this scenario seems quite plausible to me. But clearly the take-off
would lead to a wildly unpredicable and possible highly unstable
change process.

At such a time of rapid change many players will be seeking ways to
make the change process managable. The notion of "manageable
change" doesn't imply that we or in fact any advanced intelligence can
predict the actual changes or that anyone can control the change in a
determinisitic way. All that it suggests is that the players involved in
real time in the change process might have a sense that the change
process is (somewhat) manageable. (like successfully navigating
through rapids in a canoe going down an unknow river)

I think the achievement of something that could be described as
"manageable change" will only be possible if a significant % of the AGI
mental power is actually directed towards making the change process
manageable. ie. one of the chief drivers of change also needs to be an
agent of for making that change manageable. However, how the
resources could be generated to make this possible is an interesting

Cheers, Philip

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:46 MDT