One possibly very useful application of AI would be automated theorem proving. Even the fairly limited systems which we have now are starting to be seriously useful, but they are limited by the fact that the space of possible proofs is just too big and their heuristics just too weak — they are fine for verifying proofs, and they can come in handy for some specific problems (for example, combinatorial proofs in which you have to check lots of different cases, most of them straightforward, and be sure that you got all of them), but they still are way too constrained.
Give a computer human-level intelligence and intuition, however, and you'll have an excellent potential mathematician...
edited 10th Mar '12 4:39:48 PM by Carciofus
But they seem to know where they are going, the ones who walk away from Omelas.For space exploration, the lag would be a serious issue. You'd want semi-autonomous systems at least, and ideally entirely autonomous ones.
For example, I heard that NASA is very much interested in the Belief-Desire-Intention software model
, a pretty cool family of frameworks for representing and implementing (weakly) intelligent agents. That's not even near Turing-equivalence, at least not in its current forms; but still, it allows for a respectable amount of flexibility and on-the-fly planning.
AI would be used for the same reason anything robot-related is used: To avoid having to use human labor. An employer would save a ton of money if they replaced their middle management with unpaid computers.
Of course, this would totally suck for the majority of the population...
edited 10th Mar '12 6:30:08 PM by RTaco
Usefulness for AI?
Off the top of my head; much faster identification of unknown chemicals and molecular samples (against a very large reference collection, narrowing things down faster than a normal computer can by intelligent choices), much faster advancements in mathematics and theoretical sciences, faster design of engineering and experimental science apparatus, ability to do a lot more detail work on working models of building and infrastructure plans, self-driving cars, removal of ability to "hijack" an airplane (by making pilots secondary elements, in case of AI failure), increased safety of transit systems by city-wide regulation of transit and traffic lights, increased safety of air travel by networking to avoid in-air collisions...
There's a lot of them.
For robots, there's everything from simple things like the Roomba (ease of house cleaning) to the massive practicality and lowered human cost of robotic manufacturing (compare employee deaths in electronics factories in North America to those in China), to opportunities for keeping police officers and civilians safer in the events of armed gunmen or bomb threats (the latter is already in practical use), to reducing risk of our soldiers being killed, to helping get people out of emergency situations without risking other human lives (burning/collapsed buildings, car accidents, falls in remote areas, avalanches in mountains that aren't stable yet), and so on...
Really, it all depends on what kind of scale you're looking at. There are thousands of uses for both of them, though, without even attempting to get into uses for "everyone."
"Lock up your girlfriends, lock up your wives, Grim's on the loose so run for your lives." - PyriteI cannot say I am particularly happy at how military applicability is still one of the main driving forces behind innovation. It cannot be avoided for now, I guess; but still, I really hope that we outgrow all of this silliness asap. As I see it, an autonomous beer-brewing robot is not only easier to build than a military robot, but also more useful: I like good beer, but I am not really a fan of bullet woulds...
Really, it's just the prisoner's dilemma issue. Everybody would gain if military applications of modern technology were outlawed; but if only one defects, then that one gets a huge advantage and all the others get a huge disadvantage. That's not really an optimal situation.
edited 11th Mar '12 4:19:53 AM by Carciofus
But they seem to know where they are going, the ones who walk away from Omelas.

How useful would things like robots and AI be in Real Life? Robots (for the purposes of the discussion, automated machines with no real AI) wouldn't be very fast and can't really do complicated things. Sure, they could assemble things, and maybe serve as some sort of security, or, I don't know, push carts, but would it be worth the time and resources needed to build, maintain, and debug them? AI would be another thing entirely; how could they form descisions and analyze culture (like Shakespeare, or Da Vinci or Van Gogh, or even something as simple as Animorphs?) like humans?
I eat bacon sometimes.