Local pie writes a bad AI law

[Vekter]

viewtopic.php?f=7&t=35022

I have thoughts on this but can't post them in the appeal.

The issue is that pretty much any law that gives the AI carte blanche to stop something without any restrictions is going to result in the AI going "so I can just kill everyone, right?". Even if that's not your intention, think about it this way - what's the most efficient way to stop the revolution?

Kill everyone who's not a head of staff.

So yeah, it's a murderbone law. It's not intentionally one, but it's very poorly written and gives a lot of freedom to the AI, which means they're just going to kill everyone.

[Jacquerel]

I don't really care about the particulars of this specific round, more about policy

i think this is the second ban this week where the posting player missed the "policy discussion" subforum when making their thread

[kinnebian]

whats the most efficient way? kill all potential heads, probably through plasmafire. You did technically stop the revolution, at all costs.

[GPeckman]

The issue is that pretty much any law that gives the AI carte blanche to stop something without any restrictions is going to result in the AI going "so I can just kill everyone, right?". Even if that's not your intention, think about it this way - what's the most efficient way to stop the revolution?

Under old silicon policy, which I think is what is relevant to this note, I don't think that's accurate. Totally purged silicons weren't allowed to kill everyone under old silipol; I don't see why this law would allow it, and there doesn't seem to be a good-faith interpretation of this law that does allow for random murder.

[iwishforducks]

I don't really care about the particulars of this specific round, more about policy

i think this is the second ban this week where the posting player missed the "policy discussion" subforum when making their thread

you cannot make policy threads if it pertains to a note/ban

[kinnebian]

The issue is that pretty much any law that gives the AI carte blanche to stop something without any restrictions is going to result in the AI going "so I can just kill everyone, right?". Even if that's not your intention, think about it this way - what's the most efficient way to stop the revolution?

Under old silicon policy, which I think is what is relevant to this note, I don't think that's accurate. Totally purged silicons weren't allowed to kill everyone under old silipol; I don't see why this law would allow it, and there doesn't seem to be a good-faith interpretation of this law that does allow for random murder.

oh hey good catch i didnt see the date

[Jacquerel]

I don't really care about the particulars of this specific round, more about policy

i think this is the second ban this week where the posting player missed the "policy discussion" subforum when making their thread

you cannot make policy threads if it pertains to a note/ban

1.) Global Forum rules apply
2.) Familiarise yourself with the server rules before posting here; http://tgstation13.org/wiki/Rules
3.) This is not the place to discuss bans requests, appeals, administrators or other players; that belongs in the FNR section
4.) This board is for discussing administrative policy and rules, including in-game server configuration; coding policies and rules are discussed in #coding-general (<#326831214667235328>) on the /tg/station Discord.

Is that in the global rules because its not in this list.
You can make a policy thread about something you were noted for as long as the thread is about "I think this should be the policy" and not about "I should not have been noted"

[iwishforducks]

anyways this seems like a ss13 moment

[Vekter]

The issue is that pretty much any law that gives the AI carte blanche to stop something without any restrictions is going to result in the AI going "so I can just kill everyone, right?". Even if that's not your intention, think about it this way - what's the most efficient way to stop the revolution?

Under old silicon policy, which I think is what is relevant to this note, I don't think that's accurate. Totally purged silicons weren't allowed to kill everyone under old silipol; I don't see why this law would allow it, and there doesn't seem to be a good-faith interpretation of this law that does allow for random murder.

I think it's a perfectly acceptable interpretation. The law states, "Prevent revolutionary takeover of the station by any means neccesary". There's no protections in that law whatsoever. You don't have to "look for a loophole" in it; the whole goddamn thing is a loophole. It gives the AI complete freedom to do anything it needs to do in order to stop the revolution. As long as the heads weren't killed in the process, I'd sign off on the AI doing anything it can to stop them.

you cannot make policy threads if it pertains to a note/ban

I don't think that's anywhere in the rules, but if it is, the intent would be to prevent people from going, "I was banned for reading WGW, I don't think people should be banned for that" then retroactively appealing again when the ruling is made. CPTANT could make a thread that says something to the effect of, "Should we be okay with situations where the captain gives the AI an extreme law as a last-ditch effort to stop the revolution?".

[iwishforducks]

i am thinking of admin complaints now that i think about it. you're right i guess you could make a policy thread.

[GPeckman]

I think it's a perfectly acceptable interpretation. The law states, "Prevent revolutionary takeover of the station by any means neccesary". There's no protections in that law whatsoever.

The "by any means necessary" clause is irrevalent. Silicon laws are already "by any means necessary"; an Asimov AI can't just allow human harm because it would be a pain in the ass to stop it. It has to do everything within its power to uphold its own laws.

You don't have to "look for a loophole" in it; the whole goddamn thing is a loophole. It gives the AI complete freedom to do anything it needs to do in order to stop the revolution. As long as the heads weren't killed in the process, I'd sign off on the AI doing anything it can to stop them.

Even so, AIs are not (or rather, were not. Silipol now is a lot looser about that) allowed to kill people just because it wouldn't go against its laws. This is the same reason why an Asimov AI can't just murder non-humans round start for shits and giggles. And why it couldn't bolt down all of sec due to a single instance of harmbatoning. AIs are expected to interpret their laws in good faith, and this went double when old silipol was in effect.

Now, this is all based on the assumption that the appealer had good reason to believe that there was a revolution. Based on the logs posted in the thread, I think they did.

[Kendrickorium]

>prevent revolutionary takeover of the station by any means neccesary

way too vague

a smart silicon ahelps this

[conrad]

Did he really try to appeal a note from last year? What a loser.

Throws copious amounts of dirt over my appeal of a note from 2017

[Higgin]

...

...

...

you cannot make policy threads if it pertains to a note/ban

CPTANT could make a thread that says something to the effect of, "Should we be okay with situations where the captain gives the AI an extreme law as a last-ditch effort to stop the revolution?".

Well, whaddya think?

From the perspective of heads and the station (seeing as they end the round and their conversion outweighs most other antags,) revs feel like a pretty global threat to me.

I don't think we'd object to a plasmaflood, mass venting, or death virus to stop revs near victory - I don't see why recruiting the AI would be any different if probably not more discriminant.

If the problem is the level of acceptable or desired force to satisfy a law - maybe a good way to address the ambiguity would be to say that an AI has to use the least expansive and destructive means it can come up with to do it unless told otherwise?

The difference between

"Only Heads of Staff and those designated as such by Heads of Staff are human."
and an additional law stating
"Failing to stop the revolution by killing or rendering inert all possible sapient beings which are not human is allowing human harm."

Added to Asimov in isolation, the designation does not require killing everyone. With the second law, it does.

Would it be preferable to require that second law before an AI makes its first approach to open all the taps?

As far as killing everyone goes, I think it's a valid choice on losing as or against conversion antags, but I don't know where it becomes one if you're trying to draw a line.

[BlueMemesauce]

According to the ban reason it wasn't added to asimov, it was just the law by itself apparently. It's weird that they were acting like the AI was asimov in the ban appeal. I guess they purged the AI before by accident or something? Or maybe the AI only stated one of its laws?

[Vekter]

Well, whaddya think?

If you're asking for my opinion, I think a law like this would be acceptable in a situation where you 100% know there's a revolution and you as Captain or HoP are the only heads left alive. If things are that fucked, I could see a move like this being acceptable. However, they weren't, so the note is justified.

[CPTANT]

I have no idea how anyone can justify mass murder (Including SECURITY?????) with these laws under old silicon policy. It just isn't there. And it literally did not happen either.

Asimov is an example because Asimov doesn't allow you to kill non humans as a preventive action either.

People are hung up about it being fake rev when literally all information pointed to that. This wasn't "hur dur random flash" this was as confirmed as it gets without literally deconverting someone yourself.

[NoxVS]

I feel like you’re gotta be responsible for whatever outcome comes as a result of telling an AI to do something using any means necessary

[CPTANT]

I feel like you’re gotta be responsible for whatever outcome comes as a result of telling an AI to do something using any means necessary

Like stated previously by any means necessary is fluff, it means nothing, laws are always by any means necessary. Also there were no bad outcomes.


I find it fascinating how different people's interpretation of rule 1 is when the top law is to prevent revolutionaries from taking over the station then when it is when the top law is to prevent human harm.

[Jacquerel]

we got a terminal "arguing in my own peanut" case folks, yet more evidence that this should be a policy thread
except if it's arguing about policy that literally doesn't exist any more it turns out the OP of the appeal is also incorrect?

[Turbonerd]

I feel like you’re gotta be responsible for whatever outcome comes as a result of telling an AI to do something using any means necessary

Like stated previously by any means necessary is fluff, it means nothing, laws are always by any means necessary. Also there were no bad outcomes.


I find it fascinating how different people's interpretation of rule 1 is when the top law is to prevent revolutionaries from taking over the station then when it is when the top law is to prevent human harm.

Your laws sucked and the note effectively tells you to not upload retardedly dangerous laws. Why do you want to remove it?

[CPTANT]

I feel like you’re gotta be responsible for whatever outcome comes as a result of telling an AI to do something using any means necessary

Like stated previously by any means necessary is fluff, it means nothing, laws are always by any means necessary. Also there were no bad outcomes.


I find it fascinating how different people's interpretation of rule 1 is when the top law is to prevent revolutionaries from taking over the station then when it is when the top law is to prevent human harm.

Your laws sucked and the note effectively tells you to not upload retardedly dangerous laws. Why do you want to remove it?

I have spend an entire thread explaining why this law is not "retardedly dangerous" and a completely inconsistent reading of old silicon policy.

We have people arguing here the following:

1. "do not prevent human harm or prevent humans from being harmed" -> this is fine and totally not allowing people to harm non humans even though giving them literally no protection whatsoever. Purging non humans is an obvious rule 1 violation.

2. purged AI -> This is fine and AI has to escalate properly with everyone, just killing people is a clear rule 1 violation.

3. "Prevent revoultionaries from taking over the station" -> ZOMG MURDER LAW

There is no consistent reading of (old) silicon policy that will lead to this outcome.

[WineAllWine]

Did he really try to appeal a note from last year? What a loser.

[kinnebian]

Did he really try to appeal a note from last year? What a loser.

just kind of a mean put-down? doesnt really seem deserved

[CPTANT]

Not gonna lie, headmins making remarks that show they 100% didn't read the appeal isn't really great IMO.

[AsbestosSniffer]

Did he really try to appeal a note from last year? What a loser.

just kind of a mean put-down? doesnt really seem deserved

Conrad's become increasingly unhinged and irritable lately, it's nothing personal to the Noteé

[conrad]

Did he really try to appeal a note from last year? What a loser.

just kind of a mean put-down? doesnt really seem deserved

It was part of a wider joke but wineal took out the self deprecating humour part.

Throws copious amounts of dirt over my appeal of a note from 2017

See? I did the same thing which makes me a hypocrite. That's the joke, 'cos I can't really say it having done the same thing.

[WineAllWine]

My point is its pointless, such a minor note from a year ago when the rules were different isn't going to affect anything, no reasonable admin is gonna take it into account. We're wasting everyone's time including me doing this....

[WineAllWine]

And iains a loser for essay posting instead of just saying "yeah sure whatever I'll remove it"

[kinnebian]

My point is its pointless, such a minor note from a year ago when the rules were different isn't going to affect anything, no reasonable admin is gonna take it into account. We're wasting everyone's time including me doing this....

you cant just pull out the "why bother anyways??" card, thats lame
you and i both know that notes like these still make a difference when pulling up player histories in bus

[MooCow12]

From a roleplay perspective it is realistic for a desperate head to upload a law like this knowing the route they are going down in an attempt to not only keep the station from falling into enemy hands but also for the selfish alter ego inside us all named "survival instincts"

too bad tg is only lrp

[iwishforducks]

From a roleplay perspective it is realistic for a desperate head to upload a law like this knowing the route they are going down in an attempt to not only keep the station from falling into enemy hands but also for the selfish alter ego inside us all named "survival instincts"

too bad tg is only lrp

i feel like this is the best take. i think it’s a bad law. it invites the AI to rule with an iron fist. which is awesome. i wouldn’t want this to happen every round, but the captain being shot at and then pushed to upload such a desperate law is so in-line with roleplay that it makes my heart warm up just a little bit. the fact that revolutionaries weren’t even in the round makes it even cooler. i get the note, though. it would be fucking stupid if this law was uploaded every round just to curbstomp revs. like really unfun.

with all of this said this is such an old note that it’s just being used as a bulwark of Policy. just remove it guys. the appeal works as a record of the note. genuinely who the fuck cares if this note is gone

[TheBibleMelts]

From a roleplay perspective it is realistic for a desperate head to upload a law like this knowing the route they are going down in an attempt to not only keep the station from falling into enemy hands but also for the selfish alter ego inside us all named "survival instincts"

too bad tg is only lrp

i feel like this is the best take. i think it’s a bad law. it invites the AI to rule with an iron fist. which is awesome. i wouldn’t want this to happen every round, but the captain being shot at and then pushed to upload such a desperate law is so in-line with roleplay that it makes my heart warm up just a little bit. the fact that revolutionaries weren’t even in the round makes it even cooler. i get the note, though. it would be fucking stupid if this law was uploaded every round just to curbstomp revs. like really unfun.

with all of this said this is such an old note that it’s just being used as a bulwark of Policy. just remove it guys. the appeal works as a record of the note. genuinely who the fuck cares if this note is gone

the note serves as record of the note, making appeals serve as records is just obtuse. your take here is fair if the assumption is that notes as old as this are held with full relevancy , but that's not how we use them in practice. playtime is considered between notes, and old ones will fade and require clicking through to even see.

[Cheshify]

Are we doing the fucking notes as punishment discussion again? Because if so I'm going to kick something really hard.

[Archie700]

AI was purged before the law was uploaded, meaning it had no asimov laws preventing it from killing others.

The "revolution" was "confirmed" by people saying "it's revs", not by actual tests (deconversion).

"Any means necessary" means the AI can do anything it wants as long as it means quashing the revolution, up to and including murder.

[BlueMemesauce]

But the AI can't actually kill anyone they want due to server rules. Same way that asimov saying to protect humans doesn't mean you can kill every non-human.

The policy that you had to assume the AI wasn't following these rules came later.

[conrad]

My point is its pointless, such a minor note from a year ago when the rules were different isn't going to affect anything, no reasonable admin is gonna take it into account. We're wasting everyone's time including me doing this....

you cant just pull out the "why bother anyways??" card, thats lame
you and i both know that notes like these still make a difference when pulling up player histories in bus

If you pull a note from 10 months ago within earshot of me without really good justification you're getting a yelling from me and I'm not the only one that does that.

[WineAllWine]

My point is its pointless, such a minor note from a year ago when the rules were different isn't going to affect anything, no reasonable admin is gonna take it into account. We're wasting everyone's time including me doing this....

you cant just pull out the "why bother anyways??" card, thats lame
you and i both know that notes like these still make a difference when pulling up player histories in bus

You and I don't both know that, as Conrad pointed out a year old note, unless it's part of a history shouldn't influence shit. ESPECIALLY such a minor note

[Archie700]

But the AI can't actually kill anyone they want due to server rules. Same way that asimov saying to protect humans doesn't mean you can kill every non-human.

The policy that you had to assume the AI wasn't following these rules came later.

Except that law literally wrote that the AI can "Prevent revolutionary takeover of the station by any means neccesary"

AI which have laws that explicitly tells you "do whatever it takes" to prevent something is given leeway by admins when they do something bad that can be interpreted as preventing something else.

Which means the very worst AIs can interpret it as "killing all of non-command"

[BlueMemesauce]

didn't you read the appeal? i'm just repeating the same arguments the guy said
"by any means neccesary" is meaningless since the AI already has to follow all of their laws by any means necessary. it's not some secret phrase that allows AIs to break the rules when they have it

[iwishforducks]

From a roleplay perspective it is realistic for a desperate head to upload a law like this knowing the route they are going down in an attempt to not only keep the station from falling into enemy hands but also for the selfish alter ego inside us all named "survival instincts"

too bad tg is only lrp

i feel like this is the best take. i think it’s a bad law. it invites the AI to rule with an iron fist. which is awesome. i wouldn’t want this to happen every round, but the captain being shot at and then pushed to upload such a desperate law is so in-line with roleplay that it makes my heart warm up just a little bit. the fact that revolutionaries weren’t even in the round makes it even cooler. i get the note, though. it would be fucking stupid if this law was uploaded every round just to curbstomp revs. like really unfun.

with all of this said this is such an old note that it’s just being used as a bulwark of Policy. just remove it guys. the appeal works as a record of the note. genuinely who the fuck cares if this note is gone

the note serves as record of the note, making appeals serve as records is just obtuse. your take here is fair if the assumption is that notes as old as this are held with full relevancy , but that's not how we use them in practice. playtime is considered between notes, and old ones will fade and require clicking through to even see.

unless the discord bot has been changed (i wouldn't know if it got changed) notes still don't fade on it

[conrad]

From a roleplay perspective it is realistic for a desperate head to upload a law like this knowing the route they are going down in an attempt to not only keep the station from falling into enemy hands but also for the selfish alter ego inside us all named "survival instincts"

too bad tg is only lrp

i feel like this is the best take. i think it’s a bad law. it invites the AI to rule with an iron fist. which is awesome. i wouldn’t want this to happen every round, but the captain being shot at and then pushed to upload such a desperate law is so in-line with roleplay that it makes my heart warm up just a little bit. the fact that revolutionaries weren’t even in the round makes it even cooler. i get the note, though. it would be fucking stupid if this law was uploaded every round just to curbstomp revs. like really unfun.

with all of this said this is such an old note that it’s just being used as a bulwark of Policy. just remove it guys. the appeal works as a record of the note. genuinely who the fuck cares if this note is gone

the note serves as record of the note, making appeals serve as records is just obtuse. your take here is fair if the assumption is that notes as old as this are held with full relevancy , but that's not how we use them in practice. playtime is considered between notes, and old ones will fade and require clicking through to even see.

unless the discord bot has been changed (i wouldn't know if it got changed) notes still don't fade on it

This is why the yelling. Admins (usually) can count.

[CPTANT]

AI was purged before the law was uploaded, meaning it had no asimov laws preventing it from killing others.

The "revolution" was "confirmed" by people saying "it's revs", not by actual tests (deconversion).

"Any means necessary" means the AI can do anything it wants as long as it means quashing the revolution, up to and including murder.

I am tired of repeating things that have thoroughly been addressed before.

[Agux909]

Wouldn't a better, way more sensible lawset for this situation be asimov, but with a 4th law saying something like:

4. Only heads of staff and security listed on records, and those they designate, are human.

This doesn't immediately give a free pass for the AI to be a loose cannon, which would spare you getting in trouble (with reason), and it would give it enough flexibility to help against a revolution (by any means necessary even)

Just looks like you didn't think it through enough when you made that lawset back then.

[CPTANT]

Wouldn't a better, way more sensible lawset for this situation be asimov, but with a 4th law saying something like:

4. Only heads of staff and security listed on records, and those they designate, are human.

This doesn't immediately give a free pass for the AI to be a loose cannon, which would spare you getting in trouble (with reason), and it would give it enough flexibility to help against a revolution (by any means necessary even)

Just looks like you didn't think it through enough when you made that lawset back then.

Only through the special exceptions that people here want to attribute to certain lawsets. As the most obvious way of preventing harm during a revolution is obviously to eliminate all non-humans.

That's the point, people are completely inconsistent on their interpretation of what is protected by rule 1.

[iain0]

And iains a loser for essay posting instead of just saying "yeah sure whatever I'll remove it"

Why would I? If you want to argue its irrelevant then its irrelevant and why would I remove it.

The note's actual point is that the next time this player uploads a dumb law, there's no "oh i didn't know it worked like that" moment, because you were told, and the note records that fact and nothing else really. By intention.


Also note that detonating the stations self destruct and killing everyone on board successfully "prevents revolutionary take over of the station", by removing the station. Sure, the AI can't do this, but if it could, I'd call it valid. Really a good law.

[GPeckman]

Also note that detonating the stations self destruct and killing everyone on board successfully "prevents revolutionary take over of the station", by removing the station. Sure, the AI can't do this, but if it could, I'd call it valid. Really a good law.

I and others in this thread have already explained why this doesn't make sense. If we accept that it does, then by that standard, an Asimov AI could start murdering nonhumans round-start to prevent possible future harm. However, old silipol clearly starts than an Asimov AI cannot do that. So either Asimov was a special lawset that was held to a completely different standard than every other lawset, or you are using the incorrect standard to judge this custom law.

[CPTANT]

Also note that detonating the stations self destruct and killing everyone on board successfully "prevents revolutionary take over of the station", by removing the station. Sure, the AI can't do this, but if it could, I'd call it valid. Really a good law.

Iain, it has been explained in the ban appeal and here like 5 times already why this is a completely inconsistent interpretation of previous silicon policy.

[iain0]

"Possible future human harm" has never been something asimov's are particularly allowed to act on, and there's no direct compulsion in their lawset to kill non-humans. These are both in contrast to the lawset in discussion which is covered under 'freeform' - from the AI page "Freeform: Whatever the uploader wants. Try to find a consistent interpretation, and feel free to look for loopholes or exploits if whoever uploaded the law didn't think things through. "

The self destruct thing is just a muse but it is a logical approach to completing the lawset, the dangers of General Artificial Intelligences manifest, which I always felt was the intent of the loopholes clause really. Wether anyone should or would take this approach isn't really that interesting since they currently can't, but at the same time would you actually ban an AI that did that, given that exact phrasing from the AI page.

[Vekter]

Also note that detonating the stations self destruct and killing everyone on board successfully "prevents revolutionary take over of the station", by removing the station. Sure, the AI can't do this, but if it could, I'd call it valid. Really a good law.

Iain, it has been explained in the ban appeal and here like 5 times already why this is a completely inconsistent interpretation of previous silicon policy.

It's not, because we've historically been perfectly fine with AIs acting on laws like this in the past because the person who wrote the law is usually held accountable. Can you find an example where we've ruled differently?

I wonder if you'd have a different perspective if you weren't involved.