My problem with it -- one problem, anyway -- is that human feeling comes from our biology ("gut feelings" for instance actually involve the gut) and is the guidepost to appropriate behavior. This includes feelings of empathy, love, and so on.
No computer has human biology and no computer has any way to FEEL ANYTHING. Yet, to say it again: feelings are the guideposts to appropriate behavior. As humans we can (and often must) moderate our responses to feelings -- which even dogs etc. can do, actually -- but without actual feeling, a computer (i.e., a piece of hardware running some software) is as morally adrift as any psychopath.
Of course, machines can be programmed to ACT as if they had feelings (they can also be programmed to guide a small drone with a bomb or bio-weapon to attack perfectly innocent people, for instance) but you can't (nor can Microsoft or anyone else) program anything more complex than the old DOS "copy" command without getting surprised by something the program eventually does -- and that's not even to mention outright programming errors.
Are we doing the right thing by creating ever-stronger AI?
Dude, it'd be a terrible time to be named John Connor.
Just in --
China has filed charges against every Chinese citizen except CCP members.
I wish I saw AI as benign as you seem to.
My problem with it -- one problem, anyway -- is that human feeling comes from our biology ("gut feelings" for instance actually involve the gut) and is the guidepost to appropriate behavior. This includes feelings of empathy, love, and so on.
No computer has human biology and no computer has any way to FEEL ANYTHING. Yet, to say it again: feelings are the guideposts to appropriate behavior. As humans we can (and often must) moderate our responses to feelings -- which even dogs etc. can do, actually -- but without actual feeling, a computer (i.e., a piece of hardware running some software) is as morally adrift as any psychopath.
Of course, machines can be programmed to ACT as if they had feelings (they can also be programmed to guide a small drone with a bomb or bio-weapon to attack perfectly innocent people, for instance) but you can't (nor can Microsoft or anyone else) program anything more complex than the old DOS "copy" command without getting surprised by something the program eventually does -- and that's not even to mention outright programming errors.
Are we doing the right thing by creating ever-stronger AI?
I doubt it.
All watched over by machines of loving grace....