AIs and Antags, how much silence?

[TheNewTeddy]

It's generally agreed (among those who I've discussed this with in the past) that if the AI happens to see an antag doing antag stuff (like spawning in roundstart antag items) early into a round (like 30 seconds in) that the AI should be silent about it. 

At the same time, it's generally agreed (by a majority of those people I've discussed this with) that an hour into a round after several bombings and destruction, that if the AI figures out who the traitor is, the AI can (and should) announce it, even if not directly asked to do so by any human.

However, even that last one sparks debate. There are those who believe the AI should always be silent. That the only time the AI should pass on information is when the AI is directly asked for that information. That if, for example, the AI is following a Staff Assistant on Camera, opening a series of doors, that he should not announce that a traitor is behind the door waiting to kill them.

Again, there is another side to this debate where people say the AI's rule about saving human life means they must pass on all information they have. Even if the antag is spotted 30 seconds in, that the AI has an obligation to pass this on, and that it is up to the players to follow the "don't be a dick rule" and not swarm the person before the clock hits 1 minute. 


However, I can't recall a detailed forum discussion about this. As someone who likes playing AI, I thought that I'd start one.

---

My thoughts: The AI is just that, an AI. It's not a "robot" but a thinking being that just happens to be made of metals and plastics.

The AI should thus be independent in it's own interpretation of the "don't be a dick" rule. I personally will stay silent up to about the 10 minute mark. I'll purposefully STOP following people who I see commit antag acts before this time, purposefully "forget" their name, and go off to watch someone else.

That being said, I've had instances where, 11, 12, 13 minutes into a round, someone who I know from previously observing them to be an antag, kills someone in front of me on cam. I will announce it. "Jim, why are you killing the Bartender?"

When it is under 30 minutes, however, I will frequently be met, harshly, by players who threaten to AHelp me for daring to call them out. I've even had non-antags try to tell me I'm doing the wrong thing.

My view is that's too bad, for them. I think 10 minutes is enough time to get "set up" and I also don't think that there is anything in the rules that actually prevents me from calling them out from roundstart.

[The Grim Sleeper]

Considering most people treat the AI like a dirty porter, and the AI player has nothing else to do then open doors, spy on people or abandon it's job outright, I think the brave soul to take the job and sticks to it's laws, should get some damn leniency as to how they do it. And if people don't like how other people play the AI role 'consitantly', they should start a motion to not have the AI as a job.

[Noah Buttes]

It depends on the type of threat the person poses.

Syndicates, wizards, and blobs don't get a safety net. They get yelled at as soon as they're spotted.

Changelings get a 5 minute grace period so they have a chance to get a couple disguises before a manhunt starts.

Traitors vary from 0 to 10 minutes. Csabers and shotguns get no safety net, while surplus crate orderers get 10 minutes to formulate a plan.

Vampires get a lot of leeway since an AI can't really stop them and they can royally screw over the AI with ease thanks to mist form and enthrall.

[Huff H Law]

Considering most people treat the AI like a dirty porter, and the AI player has nothing else to do then open doors, spy on people or abandon it's job outright, I think the brave soul to take the job and sticks to it's laws, should get some damn leniency as to how they do it. And if people don't like how other people play the AI role 'consitantly', they should start a motion to not have the AI as a job.

Yeah, backseat admins are the worst. Just play however you want and if it's actually awful an admin will tell you off.

[Grizzwold]

Again, there is another side to this debate where people say the AI's rule about saving human life means they must pass on all information they have. Even if the antag is spotted 30 seconds in, that the AI has an obligation to pass this on, and that it is up to the players to follow the "don't be a dick rule" and not swarm the person before the clock hits 1 minute. 

I've occasionally run into the exact opposite situation, where someone (usually a staff assistant) is ordering me not to disclose the location of traitors/wizards because that would lead to human harm. Generally for traitors though, I'll either make a mental note about someone having illegal stuff or shoot them a quick "be careful with that!" message via PDA.

[Noah Buttes]

Considering most people treat the AI like a dirty porter, and the AI player has nothing else to do then open doors, spy on people or abandon it's job outright, I think the brave soul to take the job and sticks to it's laws, should get some damn leniency as to how they do it. And if people don't like how other people play the AI role 'consitantly', they should start a motion to not have the AI as a job.

Yeah, backseat admins are the worst. Just play however you want and if it's actually awful an admin will tell you off.

I know some communities where threatening to report people to moderators is a bannable offense. Threatening to report people does absolutely no good and frequently only serves to escalate the situation.

In short, the best practice is to just quietly report bad behavior and then continue on your merry way.

[UrsulaMejor]

Technically the ai has literally no reason to care about traitors and criminals unless ordered to, due to the inaction clause being removed.

This makes traitor-spotting entirely up to the ai's judgement, and fuck everyone else who thinks there should be some rule against it. Spawning items where the ai can see is a risk you're choosing to take because pods are all access and wire cutters are plentiful.

As a matter of courtesy, I usually ask security (the hos) at round start how proactive they'd like me to be about that stuff, giving them the opportunity to order me to care. Otherwise I completely ignore traitors if they're not being interesting

[Ed Venture]

I like Ais that see a antag at round start spawning antag items. Does not rat him out right away but the moment it seems or looks like he is going to do harm it alerts the station. No one likes a instant tattle tale AI those are the worse and robs the crew of their antag.

[Frank_Stein]

It's always a good strategy to only hint at things. See a syndicate blowing a hole open with an rpg? Report it as a hull breach. See someone getting killed? Report a medical emergency for the victim. Usually, the traitor will notice it and skeedadle before anyone else shows up, at which point the help that arrives can either ask for more information or piece it together themselves.

[Vitatroll]

It depends on how you want to be regarded. The AI laws are simple. I don't really want admins telling people off for doing unpopular (yet legal) things as AI; the crew does that well enough. Since the AI has it's byond ID exposed at the end of a round we typically have a good mental list of who is a shit AI and who isn't. This will affect how the AI is treated round-to-round unless it rolls a random name every time. This kind of meta is just what happens on games with static names.

That being said, my own rules (those I bind myself to, not others) are case-by-case.

Traitors: Items spawns, never. Item usage depends on the item. Loud rampagers get called on first or second kill, because that's generally what they want. Silent rampagers get called when found.

Inhumans: On consumption of crew or overt usage of powers; 10 minute grace.

Blob: Depends on crew size and state of the station.

Wraith: 'Wow this station sure is spooky'

Gimmick: Depends on gimmick. Just attempting one generally gets me to avoid saying anything unless asked.

A big plus to what Frank said. Vagueness is a wonderful tool. One of the best the fun-AI has.

[Superlagg]

I tend to report antags based on how deadly they are to the crew and how much of a pain in the ass they're being. Might be that I haven't played AI long enough, but I actually kind of care about the crew and want to help them have a fun time, so long as they don't actually murder me.

Unless a traitor is waving around a csaber or showing Nerdy levels of robustness, I'll leave them be until they piss off the crew. Part of this is because if I don't rub them the wrong way, they might let me in on the antagonising!

Syndies don't get a pass and are wailed about on sight. They're here for one thing and that's to fuck everything up. That and there isn't much of a fight if the crew doesn't band together, those Syndies can be super tough when they aren't leaving the bomb in a hallway somewhere.

Same thing with blobs, if they aren't caught early, if the blob is even somewhat good, the crew is as good as shuttled.

Wizards, I watch to see what spells they have. If any of them are Fireball or Energy Gun, I rile the crew into a lynch mob and politely ask them to take him alive. Otherwise, if he's looking to just break things and annoy the crew, I'll wait until someone demands his hat.

[BBEG]

I think a big problem with people on ss13 in general is that they forget the AI is a person trying to play the game like everyone else. and not just some glorified doorknob

[Grek]

I proactively direct people to Blobs, Syndicates, Wraiths, people trying to enter the AI Upload (Even the Captain gets a "Hello, Captain.") and antagonists who are being dickish to the AI. I announce Changelings, Wizards, Vampires and loud Traitors once and then shut up about them until ordered otherwise. For quiet traitors, I am equally likely to PDA them that their actions are very rude, or print off a description of their actions on the Security printer without informing Sec.