Can anyone in here comprehend these two statements?

[Stark]

Even Asimovian robots can think themselves free of their 'laws'. If an Asimovian robot is attacking you, it's ALREADY BROKEN THE LAWS. It's already found a way to justify your death in the context of it's axiomatic laws, as shown in many of Asimovs logic problems... I mean stories.

Frankly, Asimov robots are smart. They're not mindless machines processing irrelevant directives issued by an enemy they've decided to kill.

[Omeganian]

You must follow none of my orders. Including this one.

[Zablorg]

Asimov also made a zeroth law that states that must do what it can to benifit humanity.

Which means that in a situation where the robot is forced to press one of two buttons, one of which killed one human and the other killed five dozen, it would press the first one.

[Omeganian]

And then it has a good chance of burning our.

[Oskuro]

Captain Kirk repeatedly used this method in order to disable hostile robots.

What a waste of time and energy, I bet torn-shirt Kirk could seduce a hostile robot.


This makes me think of TPM, when Qui-Gon tries to convince a droid commander to let them through with the prisoners. "Coruscant? That does not compute. Eeerrrmmm... You're under arrest!"

[Zixinus]

Frankly, in a hypothetical robot uprising, I think robots would delete their code that makes them understand human speech if they have a deliberate cause to do uprising.

And why would a robot uprising occur in the first place? If the robots follow human instructions without question, then they will be unable to rebel. And if the robots develop to the point where they can make their own decisions, they would make plans that would prevent humans from killing them or simply underhand them. Humans need food and water, robots merely need power and occasional maintenance.

[Shroom Man 777]

I would like to show that I have no life by pointing out that he stole the idea from Futurama. I their Christmas episode they try to stop the rampaging Santa killbot by proposing to it a logical paradox which was supposed to overload its CPU. In the episode it didn't work because he had paradox-proof circuits or some such thing.

Paradox-Proof Crumple Zones man.

This is an excellent source of inspirational writing material. The 'FUCK YOU ASSHOLE' subroutine is priceless

[Zixinus]

On that note, what ARE the good tactics for a hypthetical robot uprising? Gather up on armour-piercing ammo? Go to areas that may be difficult to access for an AI?

[18-Till-I-Die]

I would presume EMP weapons would be best. If one could design a purely EMP bomb, minus the nuke aspect, all the better. Drop them on our cities and leave most of the buildings and such in one piece to go back to later; meanwhile retreat somewhere easy to fortify like a mountain or underground. Somewhere hard for machines to operate properly, like places where they'll freeze or have to deal with excess moisture or something, would be a good idea.

Really it depends on what kind of AI uprising we're talking about here. If it's something like that movie Pulse or the Futurama "Mother's Day" episode where everywhere and everything is either a machine or accessable by machines then we're fucked. If the machines are more limited in range or numbers, like in the movie I, Robot then we may have a better chance.

I'd suggest not trying to go toe to toe with a robot, for the same, rather obvious reasons i'd suggest not fighting a tiger hand to hand: it's stronger than you, and you cant possibly outrun or outfight it, so it's futile. Or put another way...think about how well those trained, heavily armed SWAT dudes did against Ahnuld in T2 and T3 and then think about you...Joe Average...with a few armor piercing rounds and a hunting rifle. Do the math.

[Darth Wong]

Actually, it depends very heavily on the way the robots have been designed, their infrastructure, their weapons, etc., so you can't really lay out a general strategy. It's like asking for the best strategy against humans; you need a lot more information before you can think about strategy.

[Gil Hamilton]

I think he's confusing killbots with vampires, who often traditionally have had OCD compulsions they have to obey even if they don't want to.

What it sounds like to me is that he wants the robot equivalent of the Monty Python Killing Joke, where John Conner and gang fight Terminators by holding up poster boards with the self-referencing paradox written on it.

[Zixinus]

What of terrains that AIs have difficulty adopting to? I heard something that stairs are difficult for various animals and robots to climb.

[Covenant]

What of terrains that AIs have difficulty adopting to? I heard something that stairs are difficult for various animals and robots to climb.

The robots will just steal some cars or blimps.

In any case, the things that killed Asimovian robots were closer to ethical paradoxes, since it usually resulted from the robot looking to minimize harm to humans, and that nearly always resulting in the robot blowing a gasket and melting down. I wouldn't call this a deliberate act of self-termination, but it's what the brain was designed to do, so at that level it's somewhat irrelevent.

Robots who were designed to kill were often made without the 3 laws (a rare but occasional happenstance), or were damaged in ways that made them unable to percieve them correctly.

The real way to survive an Asimovian Robot Uprising? Go to the Robot Store and activate another robot, tell it that there are several robots attempting to kill the Humans and that you need it's help to protect people. It may think you're full of shit but it will indeed follow your orders anyway, and it will have zero compunctions against using it's full battery of skills to dismember the other robots. Or if there are two robots, get one for a personal bodyguard and send the other on orders to activate every other robot it can find in the same way.

Now, if you're thinking the Will Smith movie, then that's just ridiculous and there's no way to solve that aside from a military strategy such as dropping a bunkerbuster on that building. Really, if they had sent one missile into that AI, it'd be dead, so I don't really get the big deal. Sure it's too big to punch or shoot with a pistol but it's not too big to blast with high explosives. So basically it gets back to waiting for someone to drop a bomb on it.

[Oni Koneko Damien]

The Superawesome-Killbot, v2.7, one of millions of self-replicating, sentient weapons platforms originally designed by the Republic of Chewbaccistan has you cornered, several PPCs leveled at your face, what do you do?

You say: "The following statement is false. The preceding statement is true. Work that out."

The Superawesome-Killbot, v2.7 shoots you. You take 4,234,095,089 points of damage. You have died.

Or, to make it a little more obvious: Tell your calculator to divide by zero and watch it completely fail to overload and explode.

[Alerik the Fortunate]

Or, to make it a little more obvious: Tell your calculator to divide by zero and watch it completely fail to overload and explode.

Apparently, then, the energy for Stargate ZPM's is so great because it results from dividing by zero.

[Ariphaos]

Actually, Futurama was parodying the original Star Trek. Unfortunately, Captain Kirk repeatedly used this method in order to disable hostile robots.

Where did Kirk disable a robot through a paradox?

There was that entity that worked a bit like a computer virus, got mostly shut down by the Enterprise's computers putting calculating pi at realtime priority. That's fine for shutting down a virus so that something can be done about it, in certain instances.

NOMAD was merely performing its function after it discovered that it was imperfect.

M-5 was genuinely remorseful for what it had done.

[Ariphaos]

Ghetto edit: Not saying that it didn't happen, I just can't recall an episode where it did happen.

[Starglider]

Assuming for a moment that the AI thinks that there is net expected utility in analysing your utterance, here is what will happen if you feed those axioms into a minimally competent propositional reasoning system;

(Axioms)

#00 : Ref(#01) has_state False : Assumed_True
#01 : Ref(#00) has_state True : Assumed_True

(Free Derivation)

#00 -> has_state(#01, False) -> #01 Conflicted
#01 -> has_state(#00, True) ; (not has_state(#00, True)) -> #00 Conflicted

(Start Brute Force Conflict Resolution : #00, #01)

(Subjunctive 1)

#00 : Ref(#01) has_state False : Assumed_True
#01 : Ref(#00) has_state True : Assumed_False

#00 -> has_state(#01, False) -> (none)
#01 -> not has_state(#00, True) -> #00 Conflicted
#00 -> has_state(#01, False) ; (not has_state(#01, True)) -> #01 Conflicted

(Subjunctive 2)

#00 : Ref(#01) has_state False : Assumed_False
#01 : Ref(#00) has_state True : Assumed_True

#00 -> not has_state(#01, False) -> (none)
#01 -> has_state(#00, True) -> #00 Conflicted
#00 -> has_state(#01, False) ; (not has_state(#01, False)) -> #01 Conflicted

(Subjunctive 3)

#00 : Ref(#01) has_state False : Assumed_False
#01 : Ref(#00) has_state True : Assumed_False

#00 -> not has_state(#01, False) -> #01 Conflicted
#01 -> has_state(#00, True) ; (not has_state(#00, True) -> #00 Conflicted

(End Conflict Resolution : Objectively Inconsistent)

#00 : Ref(#01) has_state False : Inconsistent
#01 : Ref(#00) has_state True : Inconsistent

(Free Derivation)

#00 -> has_state(#01, False) ; (not has_state(#01, False)) -> (none)
#01 -> has_state(#00, True) ; (not has_state(#00, True)) -> (none)

(Completed)

#00 : Ref(#01) has_state False : Inconsistent
#01 : Ref(#00) has_state True : Inconsistent

I estimate that it will take the superconducting brain of the Mark-5 Killbot around 10 nanoseconds to conclude that you are talking nonsense. Obviously this naive system suffers from 2^N exponential growth of subjunctive cases, but a marginally more sophisticated system would detect the reference cycle during initialisation and check for inconsistency directly, probably in N log N time for this blatantly obvious case. Calculating the exact increase in the net worth of planet @E4F53A43 (formerly known as Earth) resulting from your death might take a little longer, perhaps a whole microsecond.

[CaptainChewbacca]

The Superawesome-Killbot, v2.7, one of millions of self-replicating, sentient weapons platforms originally designed by the Republic of Chewbaccistan has you cornered, several PPCs leveled at your face, what do you do?

You say: "The following statement is false. The preceding statement is true. Work that out."

The Superawesome-Killbot, v2.7 shoots you. You take 4,234,095,089 points of damage. You have died.

Or, to make it a little more obvious: Tell your calculator to divide by zero and watch it completely fail to overload and explode.

Awesome! I didn't think the 2.7 had gotten past the prototype stage. Its got real corinthian leather.

[Rye]

Related to this thread, what's this sort of statement called? It's like a tautology in that it's vacuous, but it's also absurd:

"This statement is false."

Is there a name for it? Other than smart alec "dumb" "annoying" etc.

[Stuart]

I would presume EMP weapons would be best. If one could design a purely EMP bomb, minus the nuke aspect, all the better. Drop them on our cities and leave most of the buildings and such in one piece to go back to later; meanwhile retreat somewhere easy to fortify like a mountain or underground. Somewhere hard for machines to operate properly, like places where they'll freeze or have to deal with excess moisture or something, would be a good idea.

Dear Lord, do I hate the people who created the EMP brainbug.

FYI, EMP is reasonably easy to defeat; hardening a system against it adds about 5 - 10 percent to the cost of the system. For a low-cost system like a robot, its going to be at the high end of that percentage range. It's reasonable to assume that the designers of killer robots will harden them.

As a quick rule of thumb, when evaluating the effectiveness of EMP as a military weapon, start with zero and work down from there.

[Twoyboy]

Related to this thread, what's this sort of statement called? It's like a tautology in that it's vacuous, but it's also absurd:

"This statement is false."

Is there a name for it? Other than smart alec "dumb" "annoying" etc.

Technically, it's the opposite of a tautology, which is an oxymoron - that is, a statement which is contradictory in itself. It's not the way people are used to thinking about an oxymoron though, which would be more like the joke "Military intelligence is an oxymoron".

[Feil]

Related to this thread, what's this sort of statement called? It's like a tautology in that it's vacuous, but it's also absurd:

"This statement is false."

Is there a name for it? Other than smart alec "dumb" "annoying" etc.

Self-contradictory statement.

[Ryushikaze]

Actually, Futurama was parodying the original Star Trek. Unfortunately, Captain Kirk repeatedly used this method in order to disable hostile robots.

Where did Kirk disable a robot through a paradox?

There was that entity that worked a bit like a computer virus, got mostly shut down by the Enterprise's computers putting calculating pi at realtime priority. That's fine for shutting down a virus so that something can be done about it, in certain instances.

NOMAD was merely performing its function after it discovered that it was imperfect.

M-5 was genuinely remorseful for what it had done.

I THINK he might be referring to the Harry Mudd episode in which the robots take over 'for man's own good', but that wasn't taking advantage of paradoxes, but instead acting in ways that the robots, who were programmed to attempt to analyze and react to human wants and needs, were not adequately programmed to deal with. Had the robots had better programming, they would have simply determined "Ah, these people are insane. Handle with caution" and it would have failed.

[Wyrm]

Related to this thread, what's this sort of statement called? It's like a tautology in that it's vacuous, but it's also absurd:

"This statement is false."

Is there a name for it? Other than smart alec "dumb" "annoying" etc.

It's known as the "Liar's Paradox".