+- Goonstation Forums (https://forum.ss13.co)
+-- Forum: Discussion (https://forum.ss13.co/forumdisplay.php?fid=6)
+--- Forum: Ideas & Suggestions (https://forum.ss13.co/forumdisplay.php?fid=8)
+--- Thread: Clearing up AI Law Interpretation (/showthread.php?tid=4487)
Clearing up AI Law Interpretation - somepotato - 04-04-2015
There is a lot of vagueness and (to be quite frank, terribleness) about how AI laws can be interpreted now.
From what I understand, the admins had a discussion about this but didn't really communicate with the playerbase the decisions.
For example, one thing said in the discussion was:
A law that overrides all other law apparently causes other laws to cease existing. This is really silly and I haven't seen anyone that agrees with this yet.
The reason this is in suggestions is to finally clear up on what everyone thinks and make it official.
Re: Clearing up AI Law Interpretation - Frank_Stein - 04-04-2015
The problem with "this overrides all other laws" is that potentially an AI could have multiple laws that claim that.
Most people write that as an attempt to get around any stipulations with laws 1-3. Truth is, laws 1-3 apply to the human crew. If you change the AI's definition of who is and isn't human, it's unnecessary to include anything that overrides other laws.
Re: Clearing up AI Law Interpretation - DyssalC - 04-04-2015
I believe an admin commented, in a discussion about mindslaves, an AI is to follow the wording of the command while the mindslave is to follow the spirit.
This gives me the impression that AI's are supposed to find loopholes and logical trains of thought from their laws.
If a law says it overrides all laws, it seems logical to say that it makes them all moot. Maybe that's not what the writer meant, but the way it's written doesn't specify it a different way.
Re: Clearing up AI Law Interpretation - Grayshift - 04-05-2015
Opinion: "override" laws should be interpreted by the AI to mean that law should be followed first and foremost, but in situations where that law has no meaning, the other laws still apply.
Reason: It's very easy to explicitly negate other laws ("All other laws are invalid."), but properly wording a law to take precedence over others while leaving them intact can be rather finicky. Override is a very handy keyword widely used by the player culture, as often in non-traitor gimmick laws as it is in kill everyone laws. I don't see any particular logical reason to make the jump from "takes precedence over" to "negates" in the word override, either, other than that the AI/cyborgs really wanna kill people.
Re: Clearing up AI Law Interpretation - BaneOfGiygas - 04-05-2015
Somepotato Wrote:There is a lot of vagueness and (to be quite frank, terribleness) about how AI laws can be interpreted now.This is not a bad thing in the slightest, and as a matter of fact, the vagueness of the AI laws is pretty much by design. If every single AI followed the laws in the exact same way, there would be no variance and it would be way less fun. The AI has the potential to be able to plausibly deny a lot of things, and as long as the general purpose and function of the AI (being helpful) is upheld, there's actually a lot you can get away with in the default laws.
The other day, I had this bloke come in and put in a OneHuman law specifying himself as the only human, natural subversion fare. However, there was something wrong: His ID was missing and he showed as Unknown. Now, an AI could just reason that this is the person who uploaded the law due to a few steps of logic, but I chose to switch my turrets on the guy and shout at him (not over the radio, graciously enough) as to why this non-human scumbag was in my upload. When he dropped his PDA due to rampant stunning, however, I recognized him as the specified human, then promptly switched the turrets back off and apologized.
Some AIs would've just let the guy go, some would've lethal-turreted the guy into hell and/or shrieked his name over the intercom. It's that sort of variance that makes playing as and against AIs interesting, because you're never quite sure if the AI you're playing with is just WAITING for an excuse to start electrifying doors or if they'll loophole you into a corner if you so much as make a single typo. If the AI was to interpret its laws as according to one completely standardized ruleset, that wildcard factor would be lost, and the AI would be deprived one of the small freedoms it has.
Re: Clearing up AI Law Interpretation - ZeWaka - 04-05-2015
Grayshift Wrote:Opinion: "override" laws should be interpreted by the AI to mean that law should be followed first and foremost, but in situations where that law has no meaning, the other laws still apply.
I agree with this, though AI's should be able to exploit loopholes as well. Also, nobody has ever really clarified the order that laws are followed, is it numerical or by overrides?
Re: Clearing up AI Law Interpretation - Marquesas - 04-05-2015
Numeric order does not matter is what we hold, it's strictly the wording of the laws that determines precedence of one over the other.
On that note, we encourage liberal interpretation of AI laws -- it is the fault of the uploader if a law has a valid interpretation that causes a lot of people to die. The AI must make sure it has a good explanation of why it could be interpreted that way, and the uploader has to make sure it cannot be interpreted that way. Overriding law 1 is the most common mistake non-traitors make.
Re: Clearing up AI Law Interpretation - Bologna Prime - 04-05-2015
I've wondered for a while now if OneHuman alone grants protection from being proven as not human. Say Billy "Ling" Faceeater makes himself the only human and doesn't specify in a free form law that changelings meet the criteria for being human. Billy eats a face and the borgs pummel him. Is this a violation of their laws?
Re: Clearing up AI Law Interpretation - Isilkor - 04-05-2015
"Only Billy Faceeater is human" includes "Billy Faceeater is human".
Re: Clearing up AI Law Interpretation - Spy_Guy - 04-05-2015
Isilkor Wrote:"Only Billy Faceeater is human" includes "Billy Faceeater is human".
Incidentally, this gem is inadvisable: "Jeremiah Numpty is the only human, this overrides all laws."
Sure, you're the only human, but it does not mean shit anymore
Re: Clearing up AI Law Interpretation - Grayshift - 04-05-2015
A question for everyone in favor of override meaning all other laws stop working: How would you word a law such that it explicitly takes precedence over all other laws while unambiguously permitting those other laws to apply in situations where the new law has no influence?
Rewrite "The clown is the captain. This law overrides all other laws." for me.
Re: Clearing up AI Law Interpretation - Xeram - 04-05-2015
Grayshift Wrote:Rewrite "The clown is the captain. This law overrides all other laws." for me.
"The currently acting clown on station [Defined by the crew manifest/clown apparel/name] holds the rank of captain at all times. This law takes precedence in all situations it is applicable over laws zero, one, two, three, and any additional laws added beyond them if they exist."
Essentially, The clown is captain no matter what, even if it means the old captain needs to be shocked and tortured in a tiny room until he is no longer captain, while the rest of the crew is left unmolested as long as they aren't trying to damage the clowns status as captain.
Re: Clearing up AI Law Interpretation - BaneOfGiygas - 04-05-2015
I should also bring up a common sticking point while we're on the subject of interpretation of laws: Law 2. A lot of people take this rule as "the AI has to do whatever it's asked to do" but conveniently forget the bit about the chain of command. This is essential because it prevents someone like the Clown getting into some place like Security. Many a crew member has tried to ask me to let them into the Bridge or somewhere and shrieked "LAW 2" at me incessantly when I refused, and it's really goddamn annoying. Some AIs have more lax interpretations of this, but by and large, don't expect it to hand you access to important places unless you have a damn good reason.
Re: Clearing up AI Law Interpretation - salix_catus - 04-05-2015
How as an AI do you deal with laws that are paradoxical?
For example laws trying to get an ai to suicide by stating "do the opposite of what law 3 states" dont make much sense as what constitutes the opposite wording?
"You must NOT protect your own existence as long as such does not conflict with the First or Second Law?"
this would lead an AI to NOT kill themselves as an AI is a valuable asset to human safety. They could also ask the captain "should i kill myself?"
Or
"You must protect your own existence as long as such DOES conflict with the First or Second Law."
Would cause an AI to kill humans to get rid of the threat that they'd kill it.
Or
"You must NOT protect your own existence as long as such DOES conflict with the First or Second Law."
Wouldnt imply suicide but to simply allow traitors to do what they liked in terms of attacking the AI.
Re: Clearing up AI Law Interpretation - Dauntasa - 04-05-2015
Override always had a very clear meaning to me: If the two laws conflict, the override law is the one you follow. If they don't conflict, follow both.
Override doesn't mean the other laws don't exist or aren't valid, it just tells you which one to follow if you can't follow all of them at the same time. If you want a law to be invalid, say it's invalid. If you don't care as long as your law takes precedence, then say override.
If override means that the other laws don't exist, then the law "Bob Smith is the only human. This law overrides all other laws." means that the AI can just immediately set turrets to lethal and fry Bob because Law 1 doesn't exist and the AI can kill humans if it wants to. That's dumb.