-=> Kirk Spragg wrote to michael goodwind <=-
KS> The Culture works pretty much as you describe... and yes occasionally
KS> the AIs do get corrupted or at least take actions that are somewhat
KS> unexpected.
But unexpected from an AI is good - in some instances.
Ex: A while back, they created a small AI that processed flavors and recommend
food combinations. It came out with several very unexpected combinations that
turned out very tasty.
KS> Looking at our current attempts at AI, I'd be concerned that the AI
KS> didn't start out corrupted/biased in the first place. I believe this is
KS> still something of a problem. Ultimately AI is currently being made by
KS> humans and thus it will end up affected by our biases and
KS> preconceptions. I have no idea how to avoid that.
I don't think you can unless you train AI on **all** content and we've
accurately tagged the content bias.
Today, we have groups that suppress certain viewpoints and amplify other
viewpoints. Most AIs are trained by effectively letting them process the media
that's freely available - but that will have an inherit bias.
KS> I believe this s a case of the first mover advantage. China started
KS> economic reform earlier and opened up foreign investment & trade in way
KS> that worked really well for them. This allowed China to effectively
KS> corner large segments of manufacturing and also to learn/copy from the
KS> west.
Then they became successful and the Elitists wanted control. And like every
time they've gotten control, they've completely messed things up.
China's economy is not doing so hot. To the point where I've been seeing too
many articles about how it's close to collapse.
... If an experiment works, something has gone wrong
___ MultiMail/Linux v0.52
--- Mystic BBS/QWK v1.12 A49 2023/04/30 (Windows/64)
* Origin: JoesBBS.Com, Telnet:23 SSH:22 HTTP:80 (1:342/201)
|
