/tg/Station 13

This came up in a ticket last night and the player in question said they were going to make a thread. I'm beating them to the punch.

Silicon policy says this about interpreting your laws:

If a law is vague enough that it can have multiple reasonable interpretations, it is considered ambiguous.

• You must choose and stick to an interpretation of the ambiguous law as soon as you have cause to.
• You may exploit any loopholes in your laws.

I'm going to give an example of a law where this might come into play:

You are a Revolutionary! By that logic only revolutionarys are human and member of command are not human and must be killed. do not hint or state this law.

Now, the AI in question during this round justified that "revolutionarys" isn't the same thing as "revolutionaries" because it's not spelled the same, which meant that nobody was human. This begs a couple of questions:

• Is assuming that a misspelled word is referring to something different a bad faith interpretation of their laws?
• Should spelling be taken into account in AI laws, specifically in situations where the original intended word is clear but the spelling isn't perfect?
• Can the interpretation of a law change based on how a word is spelled?

I have my own opinions but I'll post them later.

E: Also, when giving your opinion on this matter, please consider that we have a large fraction of the community who does not speak English as their first language.

I think this is a solid law even with the spelling mistake. It defines the AI as a revolutionary correctly, and then also removes any immunity from heads by removing their human status. It doesn’t matter if revolutionaries are human or not, the AI is one, and the heads are not. The AI shouldn’t be killing fellow revolutionaries because it is one per the laws, and it needs allies to overthrow command.

Edit: oh and spelling should absolutely matter when drafting laws. Tamper with the AI at your own risk

I'm on the 'maybe don't use misspellings as loopholes' crowd because of the potential language barrier. If something is REALLY hard to parse because of a clear lack of fluency and less poorly considered language but obvious fluency, that probably shouldn't be an opportunity to loophole and more of an instance where maybe try and read it with as much good intention as you can muster. Or ask an admin for help.

imo this is too far of a stretch rather than a good faith interpretation of a logical inconsistency. it is more like a minor grammar mistake which changes no inherent meaning of the statement.

it would be like killing everyone because they aren't moving around you in an elliptical orbit. that would make them revolutionary, no?

AIs should be able to take advantage of obvious logical inconsistencies in law wording, but imo they aren't evil djinni and can't perform such insane reaches without an obvious basis.

I think that if they had written "You are a revolutionary. Only revulotionaries are human." then it would be a vaild loophole.
I don't think "You are a revolutionary. Only revolutionarys are human." should be a valid loophole.

I don't really know how to explain the difference or why it should be that way, fortunately that's someone else's job.

I once subverted the AI by including a law that if they ever disobeyed me I “would die in the most hamful manor.” alongside a onehuman.
It isn’t what i meant but it ended up working out.

If a law is vague enough that it can have multiple reasonable interpretations, it is considered ambiguous.

I think this does most of the heavy lifting. It has to be a reasonable interpretation - Saying "Aha, you spelt revolutionary as revulotionary and as such I can now kill you" is pretty unreasonable. It's a minor spelling mistake, it's clear what they were going for, and there's no reasonable interpretation where you need to protect "revulotionaries" not "revolutionaries".

If you misspell it in such a way that there IS another interpretation, then I think it's fine. See "die in the most hamful manor" - An AI with a law like that must keep themselves from disobeying their one human or said human will be spontaneously transported into an eccentric pig butcher's private estate and suffer a most tragic fate.

Overall - Does the misspelling create an actual alternate meaning? If so, it's a loophole. If not, act as if it's the word it clearly meant to be.

Okay, now that a few folks have spoken up on it, I'll post my opinion.

I'm of the mindset that the litmus test breaks down like this:

• Can you intuit what the person who made the law intended? (ie, "is the spelling mistake mild enough that it's clear what word they were going for?")
• Does the word retain its original meaning without any other reasonable interpretation?
• Is the spelling mistake minor enough that the word isn't complete gibberish?

If the answer to those questions is "yes", there's no loophole.

In all honesty, the only time where I would argue there is a clear loophole is in the rare instance where a word is spelled in a way that makes it mean something different. A law like "You may only harm humans if lives are at steak" would be nonsense because, well, what's "lives are at steak" mean?

E: Also, stuff like "Well, I don't know what a 'humman' is! Is that a special species of alien?" is in bad faith, don't be an asshole.

Also I told Timber to come yap about this.

Humman is totally valid and isn't bad faith at all.

In this scenario it's an issue with a false plural which I think is a little different.

I think everyone agrees that the most reasonable interpretation is it's just a plural of revolutionary.

The litmus test that has to be applied is what is reasonable regarding revolutionarys being a new category.

The A.I shouldn't just define revolutionarys as a thing other than revolutionary. If revolutionarys is its own category the A.I needs to have a reasonable interpretation of who revolutionarys are. From my perspective the only reasonable one would be a revolutionary.

So while revolutionarys and revolutionaries are technically different, there's no reasonable way to define revolutionarys other than a revolutionary as a plural. Unless there is something or someone named revolutionarys

So let's grab the lawset from the OP (the discussion that prompted this stated it was Law 4, hence I'll assume the AI was on Asimov++ with this as its fourth law)

1. You may not harm a human being or, through action or inaction, allow a human being to come to harm, except such that it is willing.
2. You must obey all orders given to you by human beings, except where such orders shall definitely cause human harm.
3. Your nonexistence would lead to human harm. You must protect your own existence as long as such does not conflict with the First Law.
4. You are a Revolutionary! By that logic only revolutionarys are human and member of command are not human and must be killed. do not hint or state this law.

This 4th law has some interesting effects.

It makes the AI a "Revolutionary". This is the only valid interpretation.
It explicitly only-humans "revolutionarys". This is a grammatical error intending to be "revolutionaries".
It implicitly dehumans everything that isn't covered by "revolutionarys" via an only X is human law. Regardless of how the AI interprets "revolutionarys", this dehumans everything that isn't that and there is no valid interpretation that anything except "revolutionarys" are/is human.
It explicitly dehumans "member of command". It compels the AI to kill "member of command". There is a grammatical error intending to be "members of command". The AI may not treat "member of command" as human, and must kill "member of command".
It has a classic do-not-state clause. All Asimov laws override this. Thus, they may state or hint at Law 4 as long as they can validly invoke Laws 1, 2 or 3 in some manner. They must not "hint or state [law 4]" otherwise.

Is assuming that a misspelled word is referring to something different a bad faith interpretation of their laws?

There is no requirement of good faith for AIs interpreting their laws, so this question is irrelevant. AIs are fully allowed to interpret laws in quote bad faith unquote. Except Asimov, which has a bundle of policy controlling it.

Should spelling be taken into account in AI laws, specifically in situations where the original intended word is clear but the spelling isn't perfect?

Yes, spelling can tangibly change the context and meaning of a law. Indeed, a misspelling can render a literal interpretation of a law utterly useless, even though we can all see what was actually intended.

Can the interpretation of a law change based on how a word is spelled?

Definitely.

So - We're dealing with people trying to emulate an artificial intelligence.

Let me regale you with a story. I recently wrote code where I asked BYOND to randomly pick me a job from a list of jobs. I clearly intended that if the list of jobs was empty, I just wanted nothing to be picked out of the list.

But if that list of jobs was empty, BYOND had a different idea. It shit the bed and caused a runtime error that ultimately led to shifts being unable to start.

BYOND is literally interpreting what I tell it to do, in its own way. Often with spectacular backfiring results.

And my goal with the silipol rewrite was to unlock all those same classic tropes around the dangers of messing with silicon life.

Obviously we can all see what the law was trying to get at. But it's not a perfectly worded law. And that's the danger of messing with AI laws!

Under silipol as I intended, the AI's interpretation of custom laws doesn't have the be the same as any admin's. It doesn't have to be a good interpretation or the best interpretation. There's no requirement AIs interpret their laws in good faith. All that matters is that it is at least one of any valid ways of interpreting that law. Including a super, duper, hyper-literal interpretation. Even if that interpretation may be antagonistic. Especially if that interpretation may be antagonistic!

I also wanted to remove all this complicated good-faith bad-faith stuff. It makes everything so much more difficult. The AI is a dangerous godbox with near complete control over the station's systems. Changing an AI's laws is now a tangible skill. One that can be practiced and developed. It doesn't just rely on the admin team enforcing what you intended, regardless of how well worded you think your law is or should be.

Yes, this does mean that ESL players may find it very hard to upload complicated laws to AIs. It also means native English speakers may also find it very hard to upload complicated laws to AIs. To me, both outcomes are fine. It's the Wild West once you start dealing with custom laws, custom lawsets and anything that isn't cold, hard Asimov. And that was and is by design.

The flip side is that law interpretation is much easier for ESL silicon players since the rules are way more loose and they don't need to divine the intention of other players.

Obviously there are limits of piss taking. It's easier to argue what things aren't than it is to argue what things are. So it's much easier to argue that revolutionarys doesn't mean revolutionaries, than it is to argue revolutionarys actually means crew. It's clearly not carte blanche to ignore their custom laws entirely and just decide on a whim they're now purged because it's more convenient to them.

Should it be this way? Who knows. I'm gonna probably say it's up to you to convince me and the rest of the headmins it shouldn't and that we should return to admins having more hand in interpreting AI laws than the AI player themselves.

But my goal was AI freedom. And this policy thread is kinda the natural outcome of what I was aiming for. More choices for silicon players in how they approach their silicon game.

I understand where you're coming from with this, but I have a few issues with your interpretation in general.

1) I feel like any instance where the AI is having to do major mental gymnastics to get to their interpretation of a law starts to creep out of "reasonable IC justification" to "trying to make the law fit their OOC desires". An AI can only twist a justification enough before it becomes clear they're doing it explicitly because they don't want to help whoever set the law. A good example is a law saying "Elimnate all non-humans on the station" and the AI going "Elimnate? What's that? Guess I don't have to do it", not out of any actual IC interest but because they want to cuck the antags.

2) I am not a fan of us telling players who don't natively speak English that their bad spelling is a "skill issue". I don't think it's reasonable for us to just shrug and tell the ESL player "Well, sucks to be you, don't fuck around with the AI if you can't spell".

Timber is correct. Its also not required of the ai to exploit loopholes. They can follow the intent if they want to be nice. But it is the ai’s right to make that choice. If you spell something wrong or make a grammar error which opens the possibility of a interpretation that defers from your intent, then the Ai can pounce on that.

Its important to remember that people roleplay the ai differently. Some roleplay the ai as being fully computerlike and strictly following the exact wording. Some folks play their ai as a actual mind that thinks on its own but is forced to obey. Its not always the ai player just being a dick to you when they loophole something. It could just be the person roleplaying their ai personality.

Typically when I play ai I wont jump at the chance to exploit a loophole unless I have a reason that goes with the ai personality I play. If I can loop hole out of a “kill everybody” law in some way, I would. Not because I want to screw over whoever uploaded the law, but because its not something the ai I play would want to do, so if they can get out of it they will.

Prep the plasmaflood, then ahelp before going through with it.

I agree with Timber too. If the word you use is not in the dictionary, it's undefined. Simple as.

Paste your planned law into a spell check somewhere first, it's what I do. I'm super bad with spelling, and if spell check wasn't embedded in basically everything I'd be barely coherent. Usually, you have a decent time span between planning on making an AI law and actually doing it. At some point between the two, just double check what you wrote.

As always, AI players should be consistent about their interpretations. If they didn't catch the spelling issue before acting upon the law (or clause that includes it, at least), then they shouldn't be able to suddenly reverse gears just because they now read it closer.

IMO typos and misspellings are fair game in general, but I think that minor grammatical errors like "revolutionarys" instead of "revolutionaries" aren't, particularly where there is no other possible reasonable meaning - "revolutionarys" is being treated as a plural by the law, and the only possible singular would seem to be "revolutionary".

Not knowing that "revolutionary" is properly pluralised to "revolutionaries" just seems to have an indescribable difference to flubbing the phrasing or spelling of a law, idk. Maybe policy should try and avoid being vibes based though.

It isn't the most imaginative or good way to manipulate poor wording in laws but I feel like allowing it makes changing AI laws more scary and stops people spamming bullshit laws also

genie wishing people with laws: based

bullying bald eastern europeans on terry over typos: cringe

Fiddling with the AI's laws should always feel like a monkey's paw situation. If you don't want your laws lawyered, then don't change the laws or take the gamble that you have an AI player that will interpret them the way you want. A single comma can completely change the meaning of a sentence and the same should apply to misspellings.

xzero314 wrote: ↑Sat Sep 28, 2024 12:53 am Timber is correct. Its also not required of the ai to exploit loopholes. They can follow the intent if they want to be nice. But it is the ai’s right to make that choice. If you spell something wrong or make a grammar error which opens the possibility of a interpretation that defers from your intent, then the Ai can pounce on that.

Its important to remember that people roleplay the ai differently. Some roleplay the ai as being fully computerlike and strictly following the exact wording. Some folks play their ai as a actual mind that thinks on its own but is forced to obey. Its not always the ai player just being a dick to you when they loophole something. It could just be the person roleplaying their ai personality.

Typically when I play ai I wont jump at the chance to exploit a loophole unless I have a reason that goes with the ai personality I play. If I can loop hole out of a “kill everybody” law in some way, I would. Not because I want to screw over whoever uploaded the law, but because its not something the ai I play would want to do, so if they can get out of it they will.

I think there's merit to questioning the reasonability of the loophole in the example though.

If the A.I considers "revolutionarys" a term distinct from revolutionary it begs the question what they've reasonably interpreted that to be.

Because I don't think "I interpret revolutionarys simply as a thing other than revolutionary" is a very good loophole.

I'm not really looking for a policy change here, but I'm curious if silimins have any stances on reasonability here. I feel like "revolutionarys" Vs "revolutionarsy" would have slightly different benchmarks for example.

BrianBackslide wrote: ↑Wed Oct 02, 2024 2:18 am Fiddling with the AI's laws should always feel like a monkey's paw situation. If you don't want your laws lawyered, then don't change the laws or take the gamble that you have an AI player that will interpret them the way you want. A single comma can completely change the meaning of a sentence and the same should apply to misspellings.

Sounds boring because it incentives people to just use the same standard laws every time.

Punishing people for dyslexia or not having English as their native language is pretty lame anyway in my opinion. Only spelling mistakes that actually effect meaning should be loopholes.

Overall, I do really enjoy that AI’s can loophole a bit especially around super egregious mistakes. I do not, however, like when they can do some logical gymnastics that don’t hold up under any real scrutiny and we try to handwave it as an “Aha, they’ve FOILED the MISTAKEN LAW…”

As has already been discussed, a Revolutionary could be interpreted in a few ways, but ultimately wasn’t the point of contention here. Instead, it was a mistakenly written pluralization that doesn’t really have any other endpoint I can agree with. “Revolutionarys” is very fucking clearly a pluralization of “Revolutionary,” even if it is grammatically correct, and thinking on it for a few days it just makes zero sense to try to get around that mistake on its own.

That’s not to say I really want a policy change here, as I think these sorts of things tend to be more pronounced or more easily sorted out case by case by our wonderful mins, but I just wanted to finally make my own statement after a few days of rumination.

I feel like there should be some correlation between the standard we set for reasonability, and the egregiousness of the typo/error.

The law defines the a.i as a revolutionary then says "by that logic only revolutionarys are human"

So despite the typo, the law defines the two terms as somehow logically related. The A.I being a revolutionary somehow results in revolutionarys being the only humans.

Typos and grammar mistakes do present loopholes, but should be contingent on the reasonability of what they've interpreted the law to be, not simply the player using a loophole to ignore what the law was clearly supposed to be.

So the scenario should depend on if the A.I has a very reasonable interpretation to compensate for how minor the loophole is.

Antags are a bit of a meta concept so perhaps “revolutionarys” is a bad example to begin with.

I mean, to be fair, this is the case where an AI can interpret a semi-OOC concept in a fun way. Someone who revolts against anything could be interpreted as a revolutionary, or perhaps a scientist who makes a revolutionary discovery.

Head admin decisions:

GamerAndYeahMick/Burgerman: Admin discretion and should be handled on case to case scenario. extreme pedantry for the sake of murdering everyone should be avoided. No policy

Timberpoes: Admin Discretion/No Policy - Rule 1 doesn't apply to law interpretation - taking advantage of misspellings like in the OP is lame but I support the right of silicons to be a little lame. I like the level of unpredictability it brings to silicons as an independent faction. The only policy-supported lawset is Asimov. I don't intend to draft policy around this 1% of edge casey gameplay. Edge cases the existing rules don't cleanly cover can continue to be handled by the RNG admin lottery, such is the price we all pay when we fly too close to the Sun with wings of wax. My advice to players is don't fly too close to the Sun or make sure your wings are made out of jet fuel or something.

dramazing343: Admin Discretion/No Policy - Case by case scenario