[ExI] How could you ever support an AGI?

Henrique Moraes Machado cetico.iconoclasta at gmail.com
Wed Mar 5 12:47:31 UTC 2008


John Grigg>You are badly anthropomorphizing the AGI.  It will most likely 
not have the same biological drives/wiring that you and I have.
>Where is Eliezer Yudkowsky when we need him? lol  I think the "whole 
>gradual coming into being of AGI combined with the integration of us into 
>it,"
>is actually the very unlikely scenario.  Purely AGI development will 
>definitely progress faster than the machine/biological interfaces that you 
>imagine.

We humans tend to anthropomorphizize (that's a big word!) everything. We do 
it with our cars, with our pets and with our computers (c'mon, tell me that 
you don't think your computer has feelings sometimes... :-)). Anyway, an AI 
programmed by humans would almost certainly resemble a human, since the only 
referential we have is us.

John>Upgrading animals would be a very cool thing, indeed.

I'd really really love to see that happen. Anybody working on it?

John>I would say this is a very big "if."  But some say AGI would only have 
the motivations which we program into them.

At first at least. But a self improving AI could change that.





More information about the extropy-chat mailing list