AI-boxing (was: Problems with [...])

From: Elias Sinderson (
Date: Fri Aug 26 2005 - 16:46:32 MDT

Tim Duyzer wrote:

> Perhaps it would be most effective to create a perfect simulation of
> our own world, and then [...]

Ha, thanks for the laugh!

Seriously though, we already have a perfect environment for an AGI to
learn about the real world - simply look around you. Don't dismiss this
idea so quickly, but consider the cost savings in not having to build
the simulation. . . and how those resources might be better spent
working on the /real/ problems at hand.

As far as boxing goes, how about putting it in a 'body'? Really, all
this talk about understanding the nature of 'qualia', without a single
mention of providing sensory modalities for the AGI to perceive the
world*. The sensory modalities provided would effectively throttle the
amount of stimulation the AGI receives, the amount of information it has
access to. The perceived risk of 'take off' would be relatively low
without direct access to the network, and an EMP burst would most
effectively shut the thing off if it became necessary.

Two cents,

* Not to (re)start this discussion, but do realize that, e.g., the
motion of atoms doesn't /cause/ heat, it /is/ heat... Same goes for qualia.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT