Sure. But I was thinking that AI will probably take the form of something like Bender the Robot from Futurama. Modern welding robots simply go through a preprogrammed set of movements. An AI welding robot would more or less do the same, but it would also be able to proactively recognize problems and maybe even figure out a more efficient way of doing things. Basically, an AI robot would problem solve on its own in the context of the task it was built for.
Take an AI cruise missle for example. I highly doubt one would ever question its purpose for existance and get depressed. What it would do is behave like a golden retriever. You say “go blow up this this warehouse” and off it goes looking for it.
1812 I thought Sikozu was genertically engineered, not a robot. Of course I’ve missed a few episodes. But the DRDs are another good example of AI. They seem to have their own personalities and are independent, but they are also specificly designed to perform a specific task. And they aren’t as obtrusive or creepy as Lance Heinricksons “Biship” from Aliens (whats the deal with that? was the Christopher Walkin model of “artificial person” sold out?)
Artificial Intelligence in independently mobile robots will probably be limited for a long time, as the processing needed to produce anything like human intelligence would need a large heavy computer (this might not be true for long, if Moore’s law has any validity).
So the individual robots could be drone extensions of a generalised universal mainframe AI, which has designs (recipes) for many different contingencies, and will manufacture new purpose built robots with limited adaptability. These drones will probably have limited intelligence, as well- just enough to get themselves to a place of safety when ever they are cut off from the guiding AI mainframe.
(It certainly seems likely that the world in times to come will be swimming in a sea of data- nearly all of it encyrpted.)
But what will their personalities be like-and what will their relationships be with humans?
The hardwired imperative of ‘Friendliness’ can hopefully be hardwired into such intelligent AI, and later into the smaller but perhaps equally as powerful processors of independently mobile Universal Robots. Friendly AI creation link
How long will this (as yet undefined ) ‘Friendliness’ persist, given that most robots will be designed by artificial intelligences themselves, and therefore subject to a rapid, one might say Lamarkian, evolution?
If the ones with Asimov type imperatives are at an evolutionary disadvantage compared with the ones free of restrictions on their behaviour then they will quickly be out competed.
Self designing Lamarkian AI would be quickly beyond human control if not placed under draconian restrictions, and the rate of acceleration of technological change could approach a Vingean Singularity.
(if you are looking for fictional treatment of the Singularity, most of Vernon Vinge’s stories are concerned with this one way or another.)
On the other hand tight controls on the behaviour of highly intelligent manufactured beings could cause electronic psychoses, or at the very least, lead to such cautious inaction that the robot could decide that it would be safest to disobey all orders and do nothing, thereby causing least harm.
Right, I see what you’re saying. In a sense you’re talking about machines that are intelligent in the way that a dog is. A dog can solve most, if not all, of the problems associated with being a dog. It can think, though in a way that we probably couldn’t comprehend, in a very real and meaningful sense. But it wouldn’t be what I mean when I say intelligence & AI. But we can disagree on that.
It is an interesting question as to whether universal machines or dedicated machines would dominate. I can imagine AI becoming so inexpensive that it really would be most cost effective to create blank robots that will have the specifics added on after the factory. That would make a good question for an econ. class.
First I appologise for taking so long to answer. I was busy and kinda forgot about the whole internet thing for a while. Now, to the thread itself:
First I clarification. If you mean by universal machine a kind of AI independent of the body I have to agree with you. I share your opinion that intelligence has a certain universal dimension insofar as its field of exploration is only limited by artificial restrictions posed by itself. However I do believe it is possible to create artificial restrictions to the possible actions of this intelligence.
If you mean that a universal machine is a kind of generalized do-all robot I have to disagree. Especialization pays.
I don’t think there’s much use in continuing this thread as we both have very basic differences in our points of view and an experiment to find out which one of us is right is impossible in the near future. Still, it was fun and I actually changed my mind a bit about some stuff and thank you for the opportunity.
Actually, I think we may be pretty close to agreement given what you’ve just posted. And thank you as well! It was fun to think about that stuff for a while, especially from a perspective other than my own. Best regards!
Now I think I’ll go toast a pastry.