[ExI] stealth singularity
spike at rainier66.com
spike at rainier66.com
Thu May 16 14:45:26 UTC 2019
I have long thought of the singularity as being kinda like the Spanish
Inquisition: it just shows up unexpected, as it did in Monte Python.
But what if. the more gradual singularity model prevails, analogous to the
way chess programs just kept getting better and better over the years. I
watched that one grow for fifty years. It would appear to be AI from the
point of view of a 1970 chess guy, but it isn't.
MicroSloth has been experimenting with making Word subtly modify one's
writing. It is nothing shocking: it changes the word "man" to person or
some gender-neutral version. Policeman becomes police officer, fireman
becomes fire fighter. OK.
The whole thing has my wheels spinning. I bet we extrapolate on the idea,
write software which would modify or translate a passage from politically
right to left, to prevent writers from being Twitter-banned. It would
automatically delete right wing (or left wing) memes, or switch them. That
could be a fun game.
I am not sure how it would work, but I don't think it really requires AI to
do it. The news people are going nuts over it and saying MicroSloth is
using AI, but it isn't that, any more than good chess algorithms are AI.
They are just "good" rather than intelligent. Two different things, often
conflated.
Spike's postulate: any algorithm we know how to write is not AI, by
definition.
A real AI could come into existence, then use that algorithm to cover its
tracks. It could disguise itself as a PC-editor, merely seeking to make
one's writing politically correct, to prevent triggering the masses.
Perhaps the singularity could sneak up on us such that when it arrives, no
one would notice. Oh, we could make a cubic buttload of money off of this.
spike
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20190516/421348bf/attachment.htm>
More information about the extropy-chat
mailing list