/tg/Station 13

bagil

Well there's your problem

terry is literally bagil lite and sybil isn’t as good anymore so bagil isn’t bad, or much different in comparison

the trick to being a good ai is to find the balance between actually upholding your laws and not making the game unfun for people

its incredibly easy for an ai to simply patrol the entire station, immediately call out every antag they see, bolt down doors and ruin any semblance of stealth, and imo the best ais are the ones that play less as overprotective station guardians and more as """helpful""" assistants that pay more attention to roleplaying and livening up the station than the guy emagging into engineering

unless its something massively dangerous like a cult or wizard i dont think people generally mind if the ai looks the other way when they catch a traitor doing non-harmful things or at least waits a bit before announcing it, and even if they are actively killing something its more fun for everybody involved for the ai to say "harm detected in x" instead of "oh god johnson is eswording the captain to death right now bolting down all doors"

basically, instead of playing to protect all humans, play a slightly dysfunctional ai that is required to help people, but don't be scared to do things that technically break your laws if nobody will ever know the difference, as long as its not actually hurting anybody or impacting the round

t. sybil player


edit: also my personal stance on harm is that self inflicted harm or harm that comes from threats humans actively and knowingly walk into isnt something you're obligated to prevent. if i tell somebody 'this room is on fire' and they say 'i dont care open up' then law 2, it opens up. technically this breaks law 1 but if you play ai to prevent all potential harm then you end up not letting anybody do anything, not letting security actually fight threats, and also youd have to shut down rage cages and nobody likes an ai that shuts down rage cages

Communication is key to being good at AI, as well as always having low expectations of the crew.

You didn't do anything wrong. The AI and cyborgs will inevitably come into conflict with the crew when they're Asimov. The conflict is intentional. It's not a bad thing. It's part of roleplay. It adds tension and stakes to the game.

Also, if a security officer can't get out of a bolted room they probably don't know what they're doing.

There are a few key points to playing AI well
1) Be observant. It's extremely easy for you, since your "body" is an invisible noclip ghost providing mostly xray vision and your real body is safe and sound* somewhere in space. Use audio cues to figure out if you're under attack since sound you hear is still centered on the core. If you hear a welding tool slicing someone is probably taking down an rwall into your core. Read radio and try to catch on to what people are doing, move your camera around and see what people are doing.
2) Know your tools and how to use them, and what humanoids can do to counter your tools. Hotkeys: CTRL-Click bolts a door, Alt-Click shocks a door, Shift-Click opens a door, CTRL-Shift-Click sets emergency access. You can click on any button that a camera can see to use it. Just click on it to open the AI door control panel. A bolted door can't be opened without hacking it first - if the bolt wire is known and the guy is decent that'll be about 3 seconds. A bolted, depowered door cannot be opened save destruction or RCD. You can depower a door with the door control panel or by shutting off environmental and/or main breaker on an APC. Someone with access has to swipe their ID and turn both off. Meanwhile you can just lock the panel after they swiped their ID and laugh. Learn how to use atmos/air alarms. This is both useful when malf/subverted and when trying to fix damage, say when Toxins set itself on fire yet again. The top-right quarter of the game screen is just a bunch of verbs and a good place to keep the crew monitor, always open. If people suddenly go off-sensors or start taking damage you might want to take note.
3) Be aware of the 'factions'. In a simple model the factions are crew, antags and silicons. Inherently you're neither pro-crew nor pro-antag. You're pro no harm being done. If the crew are lynching (human) antags by the dozen then you are anti crew and pro antag. Clever antags will use this understanding and pull you to their side and force you to help while keeping all their harmful shenanigans on the down-low. It's the garbage AIs that call human-harming nonhuman shitcurity on a human antag who's just stealing the blueprints because muh valids. If someone cries that you're a shitter, chances are he doesn't understand how to play with a good AI. Of course you just might be a shitter, I don't know you.
4) Decide on an approach. Aside from harm being done and people giving you orders, you don't need to do anything. You can observe, like people observe ants in a terrarium and watch them plod along, toil and labour and waste away. Or you can assume leadership because, let's face it, Command won't do it and you're better equipped to do it anyway. Tell people what to do, where to go and why to do your bidding, be an efficient station overseer if you like. Or just doorshock catgirls because they're rude to you like you'd step on an ant without noticing it.

*any player with triple digit IQ can kill you and you can do nothing to stop him if you don't have a good borg player. luckily that means you're safe from about 99% of the playerbase

Jack7D1 wrote:I bolted in the lone security officer to prevent further human harm.

Be careful with this. You can technically get bwoinked for it. Brush up on your silicon policy.

skoglol wrote:

bagil

Well there's your problem

kek

actually the most important part of being ai is shitposting by telling the crew there's a cult because poly said so and watching as they actually argue over calling the shuttle

nianjiilical wrote:actually the most important part of being ai is shitposting by telling the crew there's a cult because poly said so and watching as they actually argue over calling the shuttle

that's what i would have answered if he had asked how to have the most fun as AI
its fun to describe things happening in vague ways that will lead the crew to misinterpret them and start behaving like idiots

in rev rounds things will go to shit and there'll be nothing you can do about it in the slightest

prevent harmies where you can and listen to orders when asked but other than that don't stress yourself out because mister flashies mcflashie pants is trying to bludgeon grey mctide because he had sunglasses on

edit: tl;dr report it to security and sillycones and then move on

you have more important harmies you can be preventing like the captain getting fucking flamethrowered to death in the brig

to be not shit is to resist the urge of killing everyone, if you can manage that, you will be just fine

Lumbermancer wrote:to be not shit is to resist the urge of killing everyone, if you can manage that, you will be just fine

you just couldn’t resist your valids, could you?

i can't think of any particular ways to be a good snitchbot considering your prime directive is to preserve yourself, snitch on everyone, and ensure that nothing dangerous ever happens.

you cant

actually make a minimum effort to roleplay that you are a computer instead of just talking like a regular person

oranges wrote:you cant

play the game

wesoda25 wrote:

Lumbermancer wrote:to be not shit is to resist the urge of killing everyone, if you can manage that, you will be just fine

you just couldn’t resist your valids, could you?

It took couple hundred hours, but everyone breaks in the end.

bandit wrote:actually make a minimum effort to roleplay that you are a computer instead of just talking like a regular person

This is harder to do than you might think (at least for me), but it's really important. Gotta remember to slow down. It's not like you're in the middle of the action, you can afford to wait an extra few seconds considering what you're gonna say and how you'll say it.

Quite simple, respond to everything with "law 1 i cannot comply" while playing pokemon and bolt down upload.
Suicide if you get a stupid lawchange.

It's fukken practice, that's all.

MisterPerson wrote:

bandit wrote:actually make a minimum effort to roleplay that you are a computer instead of just talking like a regular person

This is harder to do than you might think (at least for me), but it's really important. Gotta remember to slow down. It's not like you're in the middle of the action, you can afford to wait an extra few seconds considering what you're gonna say and how you'll say it.

I find it's easier if you use dry and clinical language and de-humanise yourself and others. Like, for example, if you locate a dead body just say exactly that - a dead body, not a name or an occupation, just a cadaver. I like to go a step further and call janitorial for dead bodies, not medical, and refer to it as biological waste.