[extropy-chat] Superintelligence

[kevinfreels.com]

>By then it will be too late.
>If something so amazingly useful can be done, it will be done by someone somewhere. 

So you say. But you have no proof that this is the case. I can think of lots of useful knowledge gained from breeding a human and a chimp and as far as I know it hasn't been tried. That's simply amazing if you ask me. 

Also, I am working on the assumption that there is already sufficient technology at the time to augment human intelligence to the point where human level AI is not "so amazingly useful". The real benefits will come from software and agents with "AI capabilities". What's the amazing usefulness of creating a sentient being that can be hurt, insulted and must be compensated and/or treated like a slave? It's much more useful if you can keep computers just below the level of sentience and use them to enhance your own mind. You really need to step out of this box that assumes that is some incredible standalone technology that will just pop up out of nowhere. When the capability exists, Near AI stuff will already be integrated into nearly everything already.


  >The 'Wintermute' solution may be viable for a few years, but how long could a bunch of talking chimps keep a thousand mega->Einsteins under lock and key? Every chimp beyond fooling as well as being  incorruptible?

  Again, you are assuming that mega-Einsteins won't be the norm at the time. With thousands running around, what's one more?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20060215/7f80508b/attachment.html>