blindmapmaker asked: Robots/AI or any more or less mechanical/digital life-form. Is SF better off with using them a lot or better without them? (The Culture vs. Vorkosigan for example).I love this question
( Read more... )
Hey, thanks for the interesting analysis. Good food for thought. Not sure whether the three-part classification holds for all kinds of stories. I wouldn't think all the amoral AIs are necessarily monsters or villains for one - not even in dramatic function. And there's also the bumbling servant archetype of early SF that lies somewhere between person and machine (often some kind of idiot savant). It's become quite rare nowadays (the only example I can think of is B4 from ST Nemesis
( ... )
1000 GB, heh - on the other hand, the problem isn't how much information is stored in the brain, it's how that information is used by our brains that we can't quite figure out yet. Still, I think that if research continues, it won't always stay a mystery - and who is to say that AI needs to function exactly the same as our brains? That's one of my complaints regarding AI in most universes, they're too damn human. Of course we can't really write about something completely outside our experience/imagination, but it should be possible to think up computers that are little more different, shouldn't it
( ... )
Ah yes, optimism. Who'd have thought I'd have a little trouble getting in the mood there? Apart from everybody who knows me that is ^^;
As for AIs being too damn human, I guess neural nets are still the best bet for producing something that can adapt, learn and in a way think. Also most fictional AIs are created by humans or their bumby-headed, ridged-nosed equivalents. And there's the whole "making in your own image" thing going on. I think the best fictional take on what ubiquitous, culture-forming human-made AI (organic and electronic) I've read is in the webcomic Freefall. They start out with simple three-laws robots and it turns hilarious pretty fast. They're still recognisably human, though, only quirked beyond belief
( ... )
Hmm, there is one novel (though I'm not telling you which) that has Zakalwe, but for most of the book the character is using a different name and doesn't actually remember who he is (so... a lot like Use of Weapons), and I guess if you've read Use of Weapons first (which I hadn't, then, because I'm reading them out of order) then you'd probably think "eh, what a rehash of Use of Weapons, except this guy is a little less awful" - and then in the end it turns out to be Zakalwe, which is... either a rather pointless twist or quite poignant given the theme of the book, I can't quite decide. Anyway, it means that after UoW, either the Culture decided to continue using him as an agent or he survives by some other means and ends up fighting on their side again
( ... )
Comments 4
Reply
Reply
As for AIs being too damn human, I guess neural nets are still the best bet for producing something that can adapt, learn and in a way think. Also most fictional AIs are created by humans or their bumby-headed, ridged-nosed equivalents. And there's the whole "making in your own image" thing going on. I think the best fictional take on what ubiquitous, culture-forming human-made AI (organic and electronic) I've read is in the webcomic Freefall. They start out with simple three-laws robots and it turns hilarious pretty fast. They're still recognisably human, though, only quirked beyond belief ( ... )
Reply
Reply
Leave a comment