Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Neruz]

Asimov's 3 laws are pretty watertight as these things go. Like anything, if implimented incorrectly they can lead to disaster, but when implimented correctly there's no real room for error.

Of course the zeroth law throws everything out of whack, but then again if i recall it was only applied to show that robots, when released from the constraints of the 3 laws, can still be perfectly functioning 'people' in society. Simply releasing a robot from those constraints doesn't suddenly turn it into a rampaging killer like most Sci-Fi mewvehs would have you believe.

[Starver]

Asimov's 3 laws are pretty watertight as these things go. Like anything, if implimented incorrectly they can lead to disaster, but when implimented correctly there's no real room for error.

And understood.  It was the bane of Dr Calvin's life that other humans did not understand how their systems would behave under human instruction.  (Related, though obviously 1stLaw-less, HAL's interpretation of how best to succeed at the mission while keeping the objectives secret, leading up to the respective fates of the crew of the Oddyssey.)

Of course the zeroth law throws everything out of whack, but then again if i recall it was only applied to show that robots, when released from the constraints of the 3 laws, can still be perfectly functioning 'people' in society.

I saw the Giskard/Daneel Zeroth law incident as slightly different, though if you're talking about the Evitable Machine (I think it was) "revamped first law" version, or another incident, I must profess I can't quite remember the relevent plots.

With Giskard and Daneel, Giskard is 'broken' enough to contemplate the Zeroth law, and while he couldn't directly apply it, can just about implement it within Daneel's matrix so that Daneel can do The Good Work that Giskard couldn't quite have done.

The above from memory (a couple of decades have passed, and I may have slept since).  I stand to be corrected on various points.

[Neruz]

I was thinking of a different Zeroth law.

Yeah, Asimov's Zeroth Law, that a Robot must act first and foremost to the benefit of Humanity as a whole (thus allowing a Robot to injure individual humans if it benefits Humanity to do so) is a pretty dangerous one. Ends justify the means and all that.

[Baughn]

I'm not going to discuss AI morality further. People, read CFAI - yes, really, you'll find it interesting, and I'll haunt you forever if you don't. The author makes much better arguments than I ever could, and I don't feel any particular need to replicate them here.

Starver, you'd have a point about evolution, except that we're talking about induced evolution in a computer. Real-life evolution only incidentally creates intelligence through that turning out to be handy in a real-life environment; a computer algorithm could presumably be selecting directly for intelligence, with tests tailored specifically for what we're after.

So it'd probably work quite a bit faster than natural evolution. Which is not to say that it'd be safe; see the above CFAI link, which goes into this in detail.

And for those of you who overlooked this post...

Read CFAI. Click on this link.

[Soadreqm]

I was thinking of a different Zeroth law.

Yeah, Asimov's Zeroth Law, that a Robot must act first and foremost to the benefit of Humanity as a whole (thus allowing a Robot to injure individual humans if it benefits Humanity to do so) is a pretty dangerous one. Ends justify the means and all that.

I thought that if a robot was forced to harm humans to save the whole humanity, the stress would fry its brain and it would shut down, and not go on a murderous rampage. Do I completely misremember?

###EDIT###
Baughn, did you just link a book with the honest expectation that people will read it? I'd understand maybe an article of a few pages, or something like that, but that is a wall of text so long that the creator saw fit to add an index!

[LumenPlacidum]

The problem with time being continuous is you run into good old infinity, which breaks everything at a logical level. Modern caculus and advanced mathematics and physics gets around the infinity issue by ignoring it and thus claiming to have solved the problem, without actually addressing the problem. Continuous time raises a whole bunch of very hard-to-answer questions that require some rather silly contortions of science in order to explain away, discrete time raises a smaller number of relatively easy to answer questions that require almost no contortions to explain.

Wait, what?  Calculus is the field of mathematics that is intended to deal with things that are infinitely small, large, or many.  Infinity doesn't break things at a logical level.  It's a concept defined logically; the ambiguity surrounding it doesn't come from the logical or mathematical side of things, but rather of the theoretical application to the real world.  People ask "What's infinite?" but don't really get that it's a symbol used to describe relative concepts.  Infinity is farther, no matter how far; infinity is longer, no matter how long.  In fact, infinity is an issue with discrete models too.  They avoid the infinitesimal, but not the infinite.

There are aspects of math that can be considered "ignored" by most, since the entire structure of mathematics is built upon them, but infinity isn't one of them.  On the contrary, people actively study the depths of mathematics concealed by the single symbol of the lemniscate, expanding that concept into additional numerical systems of infinite numbers.  That these numbers are still being explored implies that we don't know everything about them, but it certainly doesn't imply ambiguity or willful ignorance.

Of course, if you're talking about the applicability of calculus or mathematics to the universe in which we exist then I certainly can't argue with you there.  There's nothing but repeated observation tying the two together.

[Baughn]

Baughn, did you just link a book with the honest expectation that people will read it? I'd understand maybe an article of a few pages, or something like that, but that is a wall of text so long that the creator saw fit to add an index!

Not really. It would be nice if you would.

Thing is, the problems and solutions we've been discussing here really do need a book to explain, and there's no point in trying with anything less. So you can choose not to read it, but then you have no right to keep talking about AI morality.

[Soadreqm]

Eh, I prefer the Socratic method of arguing about it until enlightenment occurs. I did read the first few subchapters, but then I got bored. Is there a definite moment where it "gets to the point", so to speak, or would I have to slog through the whole thing?

[Baughn]

Not really, no. There is a point, but it's got a hundred pages of supporting documentation; this isn't really something you can do quickly.

If it's TL;DR, I suggest reading the Interludes and FoF chapters first. They're short, to the point, and likely to catch your interest.

[Rowanas]

You've reminded me. What does TL;DR actually stand for?

[Vester]

Too long; didn't read

I think.

[Rowanas]

Vester, so you ever log off?

[Leafsnail]

So I found the combat in adventure mode rather confusing. I don't particuarly care about adventur mode but I think it would be interesting to know how exacly combat works for DF mode. Things like counterattacks and grips.

Yup, I think we've solved his problem entirely (in much the same way that I solve the problem of noble demands with magma).

[Vester]

Vester, so you ever log off?

Yup.

I'm a lit major. I have constant free time.

(well, not today. It's 5.45 AM right now, and I need to study for exams.)

EDIT: Wow, I looked at my profile stats and apparently I'm online practically every hour besides 2 AM to 4 AM, and not even exclusively those.

I guess it helps that this is the only forum I actually visit.

[Ironhand]

And most of your posting happens around midnight, and you make an average of 44 posts every 24 hours?

...ever think about seeing someone about that?