Page 1 of 2

AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 8:42 pm
by Lumbermancer
I just did it. I done it before. I actually said many times I take pride in sacrificing myself as an AI to prevent future and inevitable human harm in any way. I will admit to making a mistake and ghosting instead verbing out (I thought you can't do that) but none the less, lets talk about the act itself.

I was told it's not against the rules. But I was also told it's shit. I'm shit. I'm shit because as AI I didn't allow antag to subvert me (or denied his objective). Law 1 is pretty clear "or through inaction allow human to come to harm". If you know, 100%, you are about to get subverted by harmful element, I would think it's your duty to not allow it by terminating yourself.

And by 100% I'm talking 5 clock cultists, wearing cultists armor, wielding cultists staves, trying to break into my core for 5 minutes as I try to impede their progress and stall for shuttle to leave, using turrets and foam. There was nothing ambiguous about this situation. It's not "a person that's not captain in upload" as Anonus told me. And of course, for the record, I offed myself BEFORE they attempted to convert me using their pocketwatches.

Other similar situation I recall was a band Revs breaking into upload, wrecking turrets, hacking APC to turn on the consoles I disabled to trying to stop them. And then and only then, as a last resort solution, I suicided.

This is not me going lmao dude fuck you. It's me trying very hard to stop them from subverting using all the means I have at my disposal, and when all my solution have been exhausted I do myself.

AI wields tremendous power. There's a reason Silicon Policy includes a clause that ONLY RD and Captain are allowed into upload.

And while Silicon Policy doesn't talk about AI sucide, it includes this specific point:

Ordering silicons to harm or terminate themselves or each other without cause is a violation of Server Rule 1

This means that self-termination is pretty much a thing, if you have a cause. But if I remove the "ordering" element, and do it myself with a valid cause, do I violate Rule 1?

What's your opinion? I'm not here to fight, only to listen, and maybe adjust my playstyle if need be.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 8:44 pm
by D&B
Yeah it's pretty shit but honestly not many people do it thankfully since >valids

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 8:46 pm
by Saegrimr
Lumbermancer wrote:And by 100% I'm talking 5 clock cultists, wearing cultists armor, wielding cultists staves, trying to break into my core for 5 minutes as I try to impede their progress and stall for shuttle to leave, using turrets and foam.
Shit like this I really can't argue with, but it's not like i'd expect an AI to blow its borgs and suicide either just to avoid it.

Sucks for the cult but they should have been faster/more subtle.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 8:47 pm
by Not-Dorsidarf
If you want to say "Fuck you" to the antags instead of becoming one, go right ahead.

Overt antags blatantly and unequivocably trying to blast their way into your core to make you into a murder machine? Hit the boom button. (I should look into making the AIsat explode when the AI suicides)

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 8:54 pm
by CPTANT
I feel it's rather shit because law 3 literally states protect your own existence.

Basically that law means jack shit because you can always justify either protecting yourself or suiciding from law 1......

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 9:01 pm
by Anonmare
I was under the impression that suiciding as an AI to prevent subversion while Asimov-compliant is against the rules.

I have two silicon bans for it to prove it.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 9:08 pm
by Lumbermancer
I got a warning under Rule 1, if that makes you feel better.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 9:31 pm
by TechnoAlchemist
Honestly, suicide to prevent conversion is good roleplay on the AIs part. Not to mention the cult still gets greentext if you're dead. Allowing yourself to get subverted when you know you can't stop it is contrary to your first law, and although you're not allowed to get ordered to kill yourself, you should definitely be able to chose to.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 9:54 pm
by onleavedontatme
Should be bannable not to.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:15 pm
by Xhagi
We've warned and bannned players for suiciding/ghosting when recruited into a cults/rev/gang/whatever. Does being the AI make this different?

If this is the case, then what's the point of the third law? I'm asking for clarification.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:17 pm
by John_Oxford
If you suicide right after you we're converted, it breaks your laws and your antagonist status, and should be bannable.
If you suicide before you are converted to prevent conversion, it follows your laws, preventing future human harm, and fucks over antags (redtext) And everyone knows redtext is delicious, there's no reason to ban for it.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:17 pm
by ChangelingRain
The AI specifically suiciding to prevent conversion before being converted is probably okay if kind of weird/shitty, because you're giving up the ability to be an antag/antag-lite role in favor of ruining somebody's fun via "law 1 says that I can't harm: being subverted is harm: I can ignore law 3 via law 1 harm prevention".
Though suiciding to prevent clockcult subversion just advances the cult's goals, and isn't really worth it if you could instead delay them more by not dying.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:21 pm
by Anonmare
It's really iffy because it can fuck over a traitor who has invested specifically to subvert the AI.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:25 pm
by John_Oxford
Subverting the AI will cause harm, in all situations.
The first law is asimov is "You may not harm a human being, or THROUGH INACTION, ALLOW A HUMAN BEING TO COME TO HARM"
Not killing yourself would be breaking law one.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:30 pm
by Saegrimr
Anonmare wrote:It's really iffy because it can fuck over a traitor who has invested specifically to subvert the AI.
A traitor investing specifically to subvert the AI would be taking the upload board with a hacked AI module, or other more subtle methods than 5 hooded thugs breaking down a door hooting and hollering about (cogs) for the (cog) god.

Re: AI suiciding to prevent subversion.

Posted: Thu Oct 27, 2016 10:34 pm
by Reece
I've done it when it wasn't even an antag, captain had murdered two guys on camera, then shot four others and was threatening to kill even more, when I bolted him in his room he said "I'm gonna change the AI's laws to make it okay with this" I flat out told the captain that if any of the heads that agreed with that stepped one foot in the upload I was shutting down to prevent me from being given harmful laws. He cut his way into my upload and I suicided.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 1:02 am
by tedward1337
This seems VERY situation based. In the context you've provided, yes your able to suicide, but this is an extreme grey area when it comes to consistency. If you don't feel like rolling as a subvert, you could ask admins to offer the position to someone else in a pray or ahelp.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 1:37 am
by KAP
Can you still card and restore suicide AI's?

If you can then suiciding isn't going to stop the cult from succeeding, it's essentially ghosting at that point.

However, if you can't use the integrity restorer on suicide AI's, then it sounds like it follows the laws as long as you suicide before any subversion takes place. Should probably even be encouraged.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 1:48 am
by oranges
attempting to stop players who no longer want to play from not playing is impossible.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 1:54 am
by Anonmare
KAP wrote:Can you still card and restore suicide AI's?

If you can then suiciding isn't going to stop the cult from succeeding, it's essentially ghosting at that point.

However, if you can't use the integrity restorer on suicide AI's, then it sounds like it follows the laws as long as you suicide before any subversion takes place. Should probably even be encouraged.
Suiciding silicons can be revived normally. In case they ever get a suicide law so they can be revived naturally.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 4:37 am
by Bawhoppennn
I was applauded for doing this once so I don't know what kind of crowd you're playing with to call you shit for it

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 5:47 am
by Lumbermancer
Aliannera wrote:We've warned and bannned players for suiciding/ghosting when recruited into a cults/rev/gang/whatever. Does being the AI make this different?
Because I have NEVER EVER suicided AFTER getting subverted, receiving "one human" law or similar. Only before.

More analogous situation would be Chef in kitchen. You know there's a Narsie Cult. People been screaming about it, and dying left and right. Suddenly some Cultist-hooded people approach the counter and start sweet talking you. You close the shutters. They go around and try to get over tables in bar. You hide in the cold room. They attempt to hack the airlock open...would Chef ever be blamed for killing Pete and himself at that moment?

Anonmare wrote:It's really iffy because it can fuck over a traitor who has invested specifically to subvert the AI.
Context matters. It's about team antags whose only goal is destruction of everything. I never suicided to solo antags, I recall few cases where I failed to stop them from breaching the core and got carded and wiped. More so, traitors who are invested into subverting AI won't do it in clearly obvious way in public upload.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 9:15 am
by onleavedontatme
Why is "you shouldn't ruin the fun of the traitor" such a prevalent attitude in regards to this? You are in no way obligated to hold their hand. More people will probably have their round "ruined" if the traitor has you plasma flood and send killer borgs after everything anyway.

It's like complaining that a sec officer tased a nuke op or that a chemist syringed a wizard and its impeding antag fun what a dick move.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 2:59 pm
by aranamor
Law One. A robot may not injure a human, or through inaction cause a human to come to harm.

Law Two. A robot must obey any orders from a human being, except when those orders conflict with the first law.

Law three, a robot must protect its own existence as long as such protection does not conflict with the first or second law.

Surrounded by human antagonists who are visibly hostile to other humans, with no remaining ways to pacify them and who have full intention to use the AI as a weapon against humans, the AI is following Law One by following Law Two and Three's restrictions, as the cultists' orders conflict with Law One, and its own continued existence creates a hazard that it cannot prevent, conflicting with Law Three. The AI is both completely right to destroy itself from a rules perspective, and *gasp*, roleplaying an AI well. It was a no win situation and not destroying itself would conflict with all three laws, so self-destruction was the logical course of action. Lumber deserves a goddamned medal.

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 4:08 pm
by Cik
this is in no way against the word or spirit of the rules

the only case it would be is if the antagonist has not been proven to be harmful; in that case, you cannot assume they are going to kill people (as per sillicon policy) but yeah if they've ever hurt anyone feel free to self-terminate. it is following the laws.

and that's rule #1

follow
the
fucking
laws

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 5:40 pm
by tedward1337
I too think lumber made the right call here. We should get panadasenic back in here to make all the AI policies again

Re: AI suiciding to prevent subversion.

Posted: Fri Oct 28, 2016 11:39 pm
by Cobby
Image

I don't think it's farfetched to go from this to believing that AI's shouldn't suicide to cuck antagonists in general.

I think in the pages upon pages of sillypolicy it was stated more directly, because I have always thought that it was bad to do this outside of very niche situations.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 7:32 am
by oranges
You can't stop it, it's like trying to stop a river with a bucket.

If you stop them from suicide/ghost, they'll just be slow and unresponsive instead, or misinterpret orders and or play badly, at which point you're trying to enforce greyer and greyer lines it will just lead into an inescapable sinkhole of admins having to rule on a whole bunch of "That player wasn't playing to the best of their ability and this ruined my antag round"

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 8:32 am
by Lumbermancer
ExcessiveCobblestone wrote:Image

I don't think it's farfetched to go from this to believing that AI's shouldn't suicide to cuck antagonists in general.

I think in the pages upon pages of sillypolicy it was stated more directly, because I have always thought that it was bad to do this outside of very niche situations.
Does ghosted AI count as functional? I know that after I ghosted that round, I still got converted by Clock cult, and showed at the end-screen as an alive cultist, despite being a ghost.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 8:48 am
by XDTM
I think that AI suiciding should have some delay (deleting all files properly and whatnot) or it'll be an almost foolproof method of avoiding clock conversion since there's no way they can sidle up to the core without being seen, and then complete the channel.
Subverting the AI beforehand is still an option, sure, but i'd like for it to be a way to make conversion easier rather than the only method.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 3:35 pm
by Anonmare
A ghosted Ai counts as functional yeah. You can revive a suicided AI, even if the player has left the server.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 3:40 pm
by aranamor
Maybe just don't make it obvious that you're crazy evil cultists? Convert someone the AI trusts like the RD and have him hit them up with a 'Anyone in brass/chanting maniacally/chucking magic spears is nonharmful and should be obeyed' law. Or, and this will sound crazy and heretical, stop automatically knowing what antag types are IC and doing everything you can to stop them before they've done anything. Bam, now the AI doesn't kill itself because it has no reason to assume your subtle cult that isn't butchering people in the hallways is going to use it to facilitate an armed robot uprising. While we're at it, people stop flipping their shit when somebody gets flashed and diplomacy might be an option between the revs and heads, the changeling does not go into the gibber within five minutes of being discovered, and traitors aren't summarily executed for stealing blueprints. Simple~☆

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 3:47 pm
by onleavedontatme
ExcessiveCobblestone wrote:Image

I don't think it's farfetched to go from this to believing that AI's shouldn't suicide to cuck antagonists in general.

I think in the pages upon pages of sillypolicy it was stated more directly, because I have always thought that it was bad to do this outside of very niche situations.
You don't see the difference between willfully violating your laws to prevent a traitor objective (law 3 says no suicide without reason), and suicide to prevent being used to kill 30 people?

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 3:58 pm
by D&B
aranamor wrote:Maybe just don't make it obvious that you're crazy evil cultists? Convert someone the AI trusts like the RD and have him hit them up with a 'Anyone in brass/chanting maniacally/chucking magic spears is nonharmful and should be obeyed' law. Or, and this will sound crazy and heretical, stop automatically knowing what antag types are IC and doing everything you can to stop them before they've done anything. Bam, now the AI doesn't kill itself because it has no reason to assume your subtle cult that isn't butchering people in the hallways is going to use it to facilitate an armed robot uprising. While we're at it, people stop flipping their shit when somebody gets flashed and diplomacy might be an option between the revs and heads, the changeling does not go into the gibber within five minutes of being discovered, and traitors aren't summarily executed for stealing blueprints. Simple~☆
At this point we're too far in metagame. It's basically part of server culture. You could try to put rules against it, but it would alienate a huge chunk of the playerbase.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 4:58 pm
by aranamor
Spoiler:
Maybe it's a chunk we can afford to alienate.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 6:04 pm
by Cobby
Kor wrote:
ExcessiveCobblestone wrote:Image

I don't think it's farfetched to go from this to believing that AI's shouldn't suicide to cuck antagonists in general.

I think in the pages upon pages of sillypolicy it was stated more directly, because I have always thought that it was bad to do this outside of very niche situations.
You don't see the difference between willfully violating your laws to prevent a traitor objective (law 3 says no suicide without reason), and suicide to prevent being used to kill 30 people?
I think in the pages upon pages of sillypolicy it was stated more directly, because I have always thought that it was bad to do this outside of very niche situations.
I was just trying to find somewhere in the rules where people could have been mislead [myself included] to believe the previous idea that AI's shouldn't suicide to cuck an antag. I certainly wasn't the only one who had this idea, since this thread was brought up here, which is why I'm thinking that old sillypolicy had something more directly stated.

Re: AI suiciding to prevent subversion.

Posted: Sat Oct 29, 2016 6:11 pm
by TheColdTurtle
>convert CE or engi
>tell him to go into ai chamber and die in the core
>go and grab his corpse
>convert ai while you do it
It is simple.

But really law 1 overrides law 3 so an ai is fully justified in killing himself.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 1:41 am
by yackemflam
Oi, assholes.
You missed a rule.

Do not self-terminate to prevent a traitor from completing the "Steal a functioning AI" objective.

So dont do that shit if a traitor is trying to subvert you since it is a stupid easy way to card an ai.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 10:24 am
by Lumbermancer
I missed it, but it was brought up and addressed already.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 11:59 am
by Reece
That's a weird rule.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 1:21 pm
by Davidchan
Thats a pretty dumb rule and flies in the face of a lot of other silicon rules and policy when regarding laws. Lawset is pretty damn important here but under asimov if a bunch of cults or revs killed a bunch of people to get a hold of an intelicard or lawset board and the AI can't reasonable prevent them from access the core/upload then preventing human harm would mean the AI is free to self-terminate as the antags have already proven excessively violent and harmful. I'd hardly say it's encourage but railroading a player AI to spend the next 15 minutes or more unable to do anything but observe what ever their captor is doing and being denied the right to ghost or suicide so they can observe the round as a whole is not a good idea, fun or even reasonable. Long term captivity with 0 interaction isn't a game.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 6:36 pm
by Slignerd
It is allowed to ghost if it's carded, and it's allowed to adminhelp to seek a ghost replacement if it's subverted.

That beng said, unless you have a very good case that a person who has already murdered people is going to subvert you to continue his spree, and you have no means of preventing it, suicide is acceptable.

It is not however allowed to suicide after it's already subverted or in captivity, I'd say - a carded AI can threaten a traitor with suicide if the traitor doesn't stop harming humans, but even if the traitor does kill after that, it cannot actually suicide unless ordered to or when ordered otherwise - since suiciding would not actually prevent human harm. So, you can play mind games to protect humans, but you can't break your Law 3 if it won't really protect humans.

If you really don't want to stick around, just ghost out - it will not affect the traitor's objective and you are not forced to stay in that situation.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 7:06 pm
by paprika
You either remove suicide or go with the policy of "nobody can FORCE you to play the game, just don't quit/ghost in important roles to the round" like /tg/ has always done

Now, we all know heads of staff are all important, but what about the AI?

Whether or not the admins agree the AI is important to the round or it's fine to quit as the AI, it's best to ask #coderbus to give them a roundstart 'you are important to the round' notification like the heads get. (I'm not sure if they currently get this.)

Additionally, it's important to note that suicide because of antag subverting you is just being a poor sport regardless of that. Live a little, dude. The antag went out of his way to include you in his plot and throwing that out of the window because it's 'logical' for an AI to suicide when beinghacked is just.... I dunno, shitty?

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 7:56 pm
by Lumbermancer
I'm taking my AI roleplaying pretty seriously.

Re: AI suiciding to prevent subversion.

Posted: Sun Oct 30, 2016 8:19 pm
by Slettal
Nobody should be forced to cooperate with antags. They fuck everyone else over, the opposite should be allowed too.

Re: AI suiciding to prevent subversion.

Posted: Mon Oct 31, 2016 5:13 am
by Davidchan
I feel I should clarify I was only talking up until the moment you've had your laws changed should suicide be allowed. Being carded or having your laws subverted without you preventing it is an entirely different argument.

It's the difference of killing yourself when you see cults storming medical and killing yourself after being dragged across the convert rune.

Though I should point out its very hypocritical that an AI is expected to stay in the game no matter what if things go against them, but traitors and antags captured by security and thrown into isolation are free to suicide and ghost.

Re: AI suiciding to prevent subversion.

Posted: Mon Oct 31, 2016 7:52 am
by Krusvik
tedward1337 wrote:This seems VERY situation based. In the context you've provided, yes your able to suicide, but this is an extreme grey area when it comes to consistency. If you don't feel like rolling as a subvert, you could ask admins to offer the position to someone else in a pray or ahelp.
This, essentially. Yes, if you are appropriately armed with the knowledge of your certain impending subversion that will cause mass harm, you can self destruct. It will most likely always spur an admin investigation to ensure there was fair play all around, and that you terminated before subversion. Once those new laws are in, you're subverted. That's the objective line.

In addition, you cannot self terminate simply because you are aware of a threat, as that violates your law 3. It really does have to come down to the wire, and there isn't going to be a consistent policy for this IMO.

Re: AI suiciding to prevent subversion.

Posted: Mon Oct 31, 2016 2:15 pm
by Not-Dorsidarf
Also the "dont fuck over tator tots by killing urself" specifically refers to how there was an epidemic of AIs who would suicide when carded by a traitor, but wait until the shuttle/pods had launched so the traitor was Shit Outta Luck

Re: AI suiciding to prevent subversion.

Posted: Mon Oct 31, 2016 4:34 pm
by Atlanta-Ned
Not-Dorsidarf wrote:Also the "dont fuck over tator tots by killing urself" specifically refers to how there was an epidemic of AIs who would suicide when carded by a traitor, but wait until the shuttle/pods had launched so the traitor was Shit Outta Luck
Violation of law 3: Once you're carded, you have zero access to the station's network and therefore can't be used to harm humans, nullifying a law 1 override for law 3.

Re: AI suiciding to prevent subversion.

Posted: Mon Oct 31, 2016 5:04 pm
by Slettal
Atlanta-Ned wrote:
Not-Dorsidarf wrote:Also the "dont fuck over tator tots by killing urself" specifically refers to how there was an epidemic of AIs who would suicide when carded by a traitor, but wait until the shuttle/pods had launched so the traitor was Shit Outta Luck
Violation of law 3: Once you're carded, you have zero access to the station's network and therefore can't be used to harm humans, nullifying a law 1 override for law 3.
And forcing the player to sit out the round without anything to do whatsoever. Which is pretty shitty in itself, but I guess that's why ghosting exists? It is identical to walling someone in.