@ai-guy, what moral values should be programmed in? What about a rule regarding the general preservation of all human life if not for all life? I think if you programmed that into it then that would cover it all.
RE: Killer Robots: Artificial Intelligence and Human Extinction
You are viewing a single comment's thread from:
Killer Robots: Artificial Intelligence and Human Extinction
I think just programming an AI not to deal any nonreversible to the planet would be OK.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Good point, but truthfully I'm not sure.
It gets very tricky. Remember, AI will take values to their logical conclusions. If we cover all life, a rational AI may conclude that humans are very dangerous to other life forms—which we are. Oppositely, imagine if AI "generally preserved" humans. The result could be AI that over-protects humans from all risk whatsoever.
But also just look at your wording: "generally" and "preservation"
What cases does "generally" not cover? What are we trying to preserve? Life? OK here comes the over-protective robots. OK fine, lets give humans some freedom. But now there is an obvious grey area in the definition.
Please search up Bostrom's paper clip maximizer
I hope this is making some bit of sense :)
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Thank you @nettijoe96 , I'll Google up Bostrom's paper clip maximizer.
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit