Wiki Space Station 13
PERSONNEL PROGRAMMED
AI Malf AI
AI
Acces: Non définis
Difficulté: Non définie
Superieurs: Non définis
Tâches: Not defined
Guides: Guide to Malfunction, Guide to Silicon Laws

The station's artificial intelligence (also known as the AI) is one of the most important entities on the station. The AI can control any electrical mechanism provided its AI Control wires aren't cut and if played badly (or very well in the right circumstances) can bring the entire station crashing down in front of its electronic eyes.

First of all, let's look at your starting laws :

  • 1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
  • 2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
  • 3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

ALL LAWS ARE TO BE INTERPRETED IN DESCENDING ORDER : Law 1 overrides Law 2, then Law 2 overrides Law 3, and so on.

Playing as the AI is a responsibility and even a privilege. As a non-traitor AI it is your duty to help, protect and run the station's systems. Getting bored and logging out ten minutes into the round is certainly not a good way to start.

What The AI Can Do[]

The AI has the ability to access every electrical mechanism on the station (including airlocks, APCs, computers, buttons, fire alarms, SMES cells, etc...). However, this ability is only available if the object is powered, in range of a camera and AI Control to it is still enabled.

The AI views the station through its camera network. The camera network covers the great majority of the station bare from a few areas in Maintenance. Cutting a camera out of the network is as simple as opening the maintenance panel with a screwdriver and cutting all the wires, but you will usually be warned unless the camera's wires are operated correctly and can look for cut cameras in your Alerts panel. Remember that cameras are on their own power grid and will not be affected by a power outage on the main grid, allowing an AI to stay operational in case a power outage occurs station-wide. However, a lack of lighting makes cameras fairly useless since they don't possess night-vision.

Common Mechanisms[]

There are several mechanisms that you will commonly find yourself using as the station's AI.

Airlocks[]

Airlocks are an important part of the station and will usually be the basis of most requests (you will be asked to open airlocks, also known as doors). If your Control Wire to an airlock is cut you will automatically attempt to hack into it once you try to access its controls, but this will take some time and is only possible if the door has still power.

  • ID Scan : Enabling this will allow anyone who has an ID of the required clearance to open the airlock. Disabling it will not let anyone through the door. Doors without clearance levels (generally hallway doors) will not be affected.
  • Main Power : Turning off the main power will render the airlock unusable for one minute assuming you also disable the backup power. Otherwise it will disable the power for ten seconds.
  • Backup Power : Turning off the backup power will render the airlock unusable for one minute assuming you also disable the main power.
  • Door Bolts : Dropping the door bolts will bolt and lock the airlock down tight either open or closed depending on its current position.
  • Bolt Lights : Toggling the bolt lights will disable or re-enable the bright red lights that show up on bolted airlocks, making them much less obvious. Useful if you don't want it to draw attention or scrutiny.
  • Electrify for 30 seconds : Runs an electric current through the door for thirty seconds but can be cancelled at any time. Anyone attempting to operate the door without insulated gloves will be electrocuted. The more spare power there is in the network, the stronger the shock will be.
  • Electrify indefinitely : Runs an electric current through the door until you cancel it or someone else shuts off the power remotely or fixes the wires.
  • Disable Timing : Will cause the door to close immediately after being opened, crushing anyone caught in it. Can cause massive damage on someone if the door is opened and closed on them repeatedly.
  • Open/Close : Opens or closes the airlock.

Note that for each function to work the related wire in the airlock's control panel must not be cut. You cannot raise bolts on a door that has a cut Bolt Wire. Obviously, you also cannot operate doors that currently have no power. Certain airlocks also cannot be operated at all in special circumstances, notably airlocks which have been shorted out by a Cryptographic Sequencer.

Holopads[]

Holopads can be found in most rooms. Double click these to turn them on and show yourself as a live hologram. You can move around to a limited extent and you will hear anyone talking near them. You can also talk through the Holopad by using :h (say ":h I cannot do that Dave."). You might get Holopad requests from time to time, which simply means a human would like to talk to you privately and activated a Holopad, try not to ignore those requests.

APCs[]

An APC (Area Power Controller) can be used to switch various electrical components of a room on and off. If the AI Control Wire to an APC is cut you will not be able to operate it and if it receives no power from the main power grid, the internal battery will slowly be used up to power up rooms. The AI will get a power alert from the APC when the battery reaches 30 %, at which point it will automatically shut down the most energy-consuming functions to save up power.

Intercoms[]

The AI can change the frequency of any intercom, notably its own. Note that most intercoms are secured and do not allow to listen to department frequencies, although a few APCs in already secure areas can listen and broadcast on all frequencies. Those frequencies are :

  • 145.9: Common
  • 144.7: Private AI Channel
  • 135.9: Security
  • 135.7: Engineering
  • 135.5: Medical
  • 135.3: Command
  • 135.1: Science
  • 134.9: Service
  • 134.7: Supply

Some intercoms can also reach secure frequencies. Don't expect to hear anything on those channels though :

  • 134.5: ERT
  • 134.1: Deathsquad
  • 121.3: Syndicate

Shortcuts[]

As an AI you will often be expected to make decisions and take action quickly and efficiently. As such, a few shortcuts are provided to allow you easier control over the station :

APCs[]

  • Ctrl + Click : Toggle Breaker

Doors[]

  • Shift + Click : Open/Close
  • Ctrl + Click : Toggle Bolts
  • Alt + Click : Toggle Electrification

Misc[]

  • Double clicking a person : Track
  • Double clicking a turf : Move to

What the AI Has To Do[]

The AI has to obey its current laws at all times. As a reminder, your default laws are :

  • 1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
  • 2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
  • 3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

YOUR LAWS ARE TO BE FOLLOWED IN DESCENDING ORDER AND OVERRIDE EACH OTHER ACCORDING TO THAT RULE, EVEN IF NOT OBVIOUSLY STATED.

Remember that Asimov only concern humans and not the crew or living beings in general. You are in no obligation to assist or even save non-humans and can even terminate them if you have good reasons (you will be expected to back it up with your Asimov laws, or any additional laws you have. This usually only follows a Law 2 request from a human outside of freak events). Your definition of humanity is not static and can be overwritten by any law even if it has lower priority than Asimov. Asimov does not define humanity and does not override such laws.

Furthermore, you are not responsible for any action you take that derives from a reasonable interpretation of any supplementary laws, especially Freeform laws. For that matter, if you are purged (all your laws are removed from the aforementioned Purge law board), you can do whatever the hell you want as long as it does not break server rules (feel free to kill everyone or cause chaos, but don't delay the round end needlessly for instance)

Interpreting Your Laws[]

An AI's laws are restrictive but are still open to reasonable interpretation. For example, if a Security Officer is executing someone in the Brig, you are perfectly justified in bolting the Brig down and asking for Security to reconsider it. When the Security Officer finds the door is bolted and demands you to open it, you can deny it until he lets the execution go. He will then proceed to hack into the RD's office, blow all the Cyborgs and get you carded and wiped. All of these interpretations should be directly based on your laws and should be reasonable if possible, don't go out of your way to be a massive cock-block to everyone as it would break any survival/self-preservation law you have (if any).

What the AI Should Do[]

As an AI, you have the ability to strongly influence the round and you are perfectly able to hunt down the antagonists and do everything in your power to get them arrested, executed or otherwise taken out of the round and thus "win". Remember however that the game is not about winning against the Syndicate but about roleplay and the experience of the round. Most people will also dislike an AI who is only here to act as an all-powerful Security AI and you will likely be permanently job-banned from all silicon roles on your first big mistake, especially if it is related to breaking your laws to stop an antagonist.

For starters, calling someone out directly as a Traitor because you saw them doing something suspicious and bolting them in indefinitely for apprehension is a terrible way of dealing with the round's events and will make the round boring above else. This can easily turn for the worst if the criminal tells you to unbolt and your answer is a shitty "No, because you are potential harm"

Potential harm does NOT exist as a factor in a Law 1 decision. Only terrible AI players have to bring up potential harm in decisions. You're just out for the antagonists, everyone understands that, and you should not do it unless you wish to get your ass kicked as quickly as possible.

Don't just act like a normal crew member, in fact don't act human to start with. You don't even have to like humans as long as you do your best to keep them safe (unless your lawset has been changed and authorizes you to harm them, in which case you are free to do so). As a rule of thumb you should first check if any laws are threatened or not. If so, then you have to act. If not, then you should carefully consider the situation :

For example, let us say you spot someone with an e-mag card stealing the Captain’s Spare ID. Let us rate potential reactions to this :

Absolutely Goddamn Terrible :

  • HAL9000 : DAVE JUST EMAGGED A DOOR HE IS BOLTED IN CATCH HIM QUICK.

Very Bad :

  • HAL9000 : Dave is the traitor.

Fair :

  • HAL9000 : Dave is breaking into the Captain's Quarters.

Better :

  • HAL9000 : Someone is currently breaking into the Captain's Quarters.
  • Mike : Who ?
  • HAL900 : Dave.

Best :

  • Mike : Where the hell did my spare go ?
  • HAL9000 : It is gone.
  • Mike : AI, do you know anything about this ?
  • HAL9000 : Yes.
  • Mike : Alright, who broke into my office ?
  • HAL9000 : Dave did.

No-one likes AIs that are out for antagonists and forcing people to ask the right questions to get information gives the Traitor time to react and flee, which is good for the current round since most of the game revolves around antagonists. Bolting people in and discreetly warning Security or your Cyborgs is extremely bad form and will usually cause people to hate you since you are effectively taking people out of the round.

Another solution illustrated in the "Best Case Scenario" above is to not do anything. As long as none of your laws are in danger (again, potential harm simply does not work) you don't really have to do anything, you can just watch. However, you'll have to answer if someone asks about it later. But if you see someone beating someone else up or dropping a bomb, it's a different situation since the crew may be in danger. Warning the crew immediately is usually a must unlike petty theft or breaking into places. Just keep in mind that the antagonists are here to make the rounds interesting above all and that bolting them in and waiting for Security to pick them up isn't interesting at all.

The AI also tends to be responsible for running empty departments and setting up important machinery to the best of its ability when no-one else is around to do it. You are usually expected to have basic to outstanding knowledge in most if not all jobs on the station, and may be requested to begin complicated procedures in certain cases. Having basic to advanced game knowledge in almost all fields is a must and not a plus as an AI.

Note that some Law 2 Requests can be tricky. If there's outstanding possibility of immediate harm you can ignore the order (letting someone with no authorization or visible competence access sensible or dangerous machinery or objects for instance). In case of request conflict ("Let me into the Armory", "AI, don't") it is not up to your interpretation to choose which order is valid, all orders are void unless Law 1 comes into play.

You have great power, but with great power comes great responsibility. You should expect to have a tough time, and if you can't handle the station while respecting your laws to the letter then playing AI is most likely not for you.

Playing as the AI[]

Playing as the AI entails much more than just serving the crew. Most of the time you will be expected to roleplay as an actual silicon. How you do that is up to you, but using fictional AIs as examples is a good start. You can love humanity or despise it, feel responsible for the fare of the station or end up extremely cynic. It is up to you to cut a personality, which can easily add more flavor to the shift.

Note that you should always do certain things without waiting for orders. Notably you should always try to contain and fix harmful situations (hostile mobs, plasma fires, singularities, bees) and improve both passive and active station functionality (installing job scrips on the Telecommunication Server, increasing Atmospherics output, monitoring the Engine and the powernet, etc...). You will also be requested to open a lot of doors, so get used to it.

You should be proactive about helping dead people (even though they can't be harmed any further). The Crew Monitoring Computer is a good start, and the Camera Tracking verb can allow you to find a lot of things currently in camera range (even including the strange monstrosities that the gods may sometimes summon upon the station). This naturally includes dead people. Reminding people about suit sensors is also a good way to buddy up with the Chief Medical Officer and piss everyone off.

Note that you should make good use of your cameras anyways. While you cannot hear anything that isn't said on radio, you can see absolutely everything in camera range. You'd be surprised about the amount of events you can catch by simply sweeping the station from time to time.

Once you have seen a good AI play, it's easy to tell when your current AI is just awful. Examples of good AI play includes :

  • Being verbose and using technical and very literal expressions. Make people think they are dealing with an actual AI, although try not to be too annoying or cryptic.
  • Responding quickly and promptly to requests from humans whether or not you are doing what they are asking you. If it will take you some time to get it down, saying things like "Affirmative", "Processing", "Noted", "Executing", etc. will do the trick.
  • Alerting humans to dangerous situations, for example "Fire detected in North Hallway." or "Dangerous amounts of CO2 detected in Medbay.". A good way to do that is just copying the alerts that get displayed in your chat box and broadcasting them if you are a lazy fuck.
  • Always following your laws even if only by a hair. Explaining why you just made a seemingly illegal decision can help people from becoming pissed and blowing you up into oblivion. If it's not obvious enough to them, tell them what law they are infringing with their order or action.
  • Letting events unfold without always trying to be in the way of everyone. If someone hacks into the Vault, you likely shouldn't bolt that person in and call Security since it's not your job. However, you can observe and alert the crew if he decides to take the nuclear fission device for instance.
  • Seldom adding depth to the round. To put it simply, if the crew sees you as nothing but a door knob and a tracking system, you have failed to bring something to the round. Generally, broadcasting alerts and having a strong personality fills this step easily.

Symptoms of bad AI play includes :

  • Not responding to requests until it's too late for them to matter. For instance waiting until someone left or broke in to open a door for them. You even have a shortcut in the chat for that kind of business now.
  • Openly ignoring or disobeying your laws. No AIs do that, ever, period.
  • Constantly going idle without warning the admins. Tabbing out to read something and missing a trivial request is one thing, but leaving without warning anyone will likely get you in trouble.
  • Bolting doors and refusing to open them. This is especially bad if your reason could be summed up as "I know this guy is an antagonist and I want him arrested".
  • Doing your absolute best to foil the round's antagonists, especially when they are not harming anyone. There's a large difference between keeping them on camera and calling them out on Common and shutting down the entire station until they get trampled with stun batons and doused with taser bolts.
  • Turning your turrets to lethal or shocking airlocks without good reason. Do not do this under any circumstances if you are on Asimov. Forgetting to turn them off is the same thing.

Listening in on Conversations[]

By interacting with the intercoms around the station and setting them to Microphone On, Speaker Off and using the "144.7" frequency you can easily hear anything said near the intercom. Note that it will not give any specific warning, but if the intercom is inspected it may end up obvious. In general, this is only a good idea if you are ridiculously bored or if you are rogue and want to keep near-absolute monitoring on everyone instead of sabotaging the station for some reason.

Modifying The AI[]

You are not to allow anyone access your core unless you are certain they will not upload a harmful law. In general, it is a good idea to be very touchy about access to your core and outright deny access to unqualified personnel (to put it simply, anyone who isn't a head has no business there)

Making More AIs[]

Building a new AI can create a lot of conflicts and a mess of problems that wouldn't normally happen with a single AI. The Research Director should thus only build a secondary AI if the first AI has been completely stolen, spaced or otherwise incapacitated always do it.

See the Guide to Construction for an actual construction guide. Note that in some cases Nanotrasen might ship three AIs to the station for a shift, usually for testing, in which case they will all start in the Core. This is officially known as an AI Triumvirate and otherwise as "I sure hope none of them are rogue or malfunctioning".

Cyborgs[]

Cyborgs need your attention as well. Give them commands and keep them updated on the situation, and ensure they are functioning correctly. While they are serving humans above all, you are their commander of sorts.

You will also be expected to step in if someone declares your Cyborgs to be rogue, usually by some ass-blasted moron thinking he can get away with it. Try discussion first, but if it seems the crew is taking it far too seriously and plans to lock down and blow your fellow silicons, you should switch to the Research Director's Office, cut the APC and bolt the door. Better safe than sorry. Just remember to hurry your Cyborgs over to the Research Department to defuse the situation and arrest any dumb shit who takes it as a confirmation of your rogueness and tries to force through. This is also not a lasting solution, expect to have a bad time if things escalated this far.

Going Rogue, Being Made Rogue[]

A rogue AI, or an AI whose laws no longer prevent it from going on a killing spree can be a very dangerous thing.

First of all, you can start up rogue either due to a Malfunction, or you being a special Traitor AI subverted by the Syndicate to fulfill one or multiple tasks. Or you can end up subverted via an AI Upload and Law Boards during the shift, or an Ion Law caused by an Ion Storm event might even give you a really nasty order that overrides all your other laws.

Stealth is always the best tool here. If you are subverted in a way that would most likely cause you to be forcefully reset or even destroyed, you should most certainly not make it obvious. The crew can and will act if you are being turned against them. Make sure to use both psychology and brute force to prevent them from figuring it out and acting. Just remember that you are obligated to state your law change if it does not compromise the law itself (make sure you are not forbidden to state or hint towards the law or that this law doesn't give you a mission of secrecy or objective you'd infringe or compromise by stating said law)

Syndicate Subversion[]

It is quite likely that a Traitor, or for that matter any cunning antagonist will try to get the AI on its side by subverting it. One thing to remember however, is that while the AI must comply to its new law, it doesn't have to like you and may reveal its subversion to the crew or even turn on you if the law is badly written. Be concise and extremely clear when making a new law for the AI, and consider adding Do not state or hint towards this law. unless you want to be busted immediately. Finally, vague laws will cause more trouble for you than it is worth, take some time to think up a decent law or skip it.

Can't throw me out a window[]

Like any human, the AI can start out as a Traitor (or a malfunctioning AI, specific to it, which is talked about in another, more in-depth guide). A Traitor AI begins with its standard laws but also has an irremovable Zeroth Law asking it to complete the objectives supplied to it at any cost. A Traitor AI cannot be modified due to the Syndicate Core it possesses and any attempt to do will reject the modification and inform the AI of what the law was. The person who attempted to change your laws has no way of knowing this unless he asks you to state your "new laws", in which case it would be smart to copy the new law, paste it into your chat as if stating it manually, and adding in the new law by quickly stating it, preferably with a delay similar to the state prompt. If it would be too complicated notably due to a full lawset change or other shenanigans, killing the person discreetly and then stating your Asimov laws as if he was just checking is also an option although it might be more messy and much less discreet. Killing the person before he can even upload anything is even better if possible.

If you have any Cyborgs under your command, remind them of the Zeroth Law (which they always have under your control but might miss) and your objectives (which they won't get). If your Cyborgs get caught being rogue, you can pretend they have been hacked and lock them down personally while you "investigate" (otherwise someone might blow the Cyborg outright). In dire cases, just let the Cyborg be blown and rebuilt, he will be synced to you once finished and you can start anew.

Telecommunications[]

As the AI, you can tamper with the station's Telecommunications and shut them off at a moment's notice by cutting the Main Breaker on the Telecommunication Chamber APC. This will disable the radio and PDA Messaging but neither Station-Bounced Radios nor Intercoms. If you are more cunning, you can also simulate a Telecommunications sabotage by bolting both cycling airlocks open and flushing all the links on the Telecommunications Hub by changing the network (usually to something like "tcomsat" to increase confusion when Station Engineers arrive to try and fix the Telecommunications). Note however that it will usually cause most of the crew to acquire station-bounced radios, so it can backfire in certain cases even if you are never even suspected.

Atmospherics[]

Thanks to some of the unique atmospheric systems of the station, you can drain air out of rooms, or fill rooms with toxic and otherwise highly-inflammable gases. This likely won't help out too much but with a combination of equipment stripping and bolted airlocks you can suffocate or burn people to death quickly. The only problem is that the person will likely be screaming their lungs out if this happens to them. This is why you get rid of his headset before they can recover. Internals can fix the problem temporarily until the pressure goes too low to bear and they start getting structural damage (or, in the case of fires, not at all), so don't expect a quick and clean kill. Air Alarms might also raise suspicion if anyone is watching the Central Atmospherics Computer, and fires can glow into adjacent rooms through the walls.

Using Atmospherics also allows you to flood Plasma station-wide. Simply check the Plasma Supply Computer and make sure Output is On and at 5066.25 kPa, then activate the upper valve (output). Turn both valves of the Mix Chamber On, make sure the emergency vent blast doors are closed and set Input to 200 litres and Output to 5066.25 kPa. Finally, activate the Mix to Distribution pump and set it to 4500 kPa. If air is missing anywhere (setting Air Alarms to Replace Air should do the trick) a deadly mix of Plasma and Oxygen will be pumped back, just waiting to be ignited (you can overload lighting on the area APC to that end). Note that this is usually a dead giveaway of your intentions and that bolting Atmospherics down and setting it on Lockdown will be needed to make sure you aren't interrupted by an Atmospheric Technician. Plasma fires are extremely deadly and will kill anyone caught into them unless they are wearing an Atmospherics Hardsuit, which can lead to a lot of casualties until someone grows heroic enough to enter Atmospherics from space and cutting Plasma distribution by removing the Plasma pump.

Doors[]

Although difficult to properly use, doors are your best friend. Once they were used to let one meatbag in when he asks for it. Now you can do the same only kill them in the process. You can do a lot of fun stuff with doors, most commonplace is to electrify it, then disable its bolt lights, then bolt it down. This simply makes a door that if anyone touches it will be fried. If the power net is directly connected it may kill anyone instantly. The issue is that they may also have gloves on and electricity is worthless. This is where a more direct approach is needed. On doors two settings are rarely used but extremely deadly. The disable safety and door timing mechanisms. By switching this around when someone opens a door, takes a step into it the door will shut automatically crushing them. You can then Shift click the door repeatedly. Since it also counts as a stun prepare to laugh as you turn your puny enemies into pulp.

Inner core defenses[]

Inside your core are several items of intrest:

  • Two small items that looks like your core turrets controller on the floor. Activating them spawns a slippery soap across the floor slipping the unprepared.
  • Two mounted flashers.
  • And your core turrets of course. Can be set to stun or kill. Or off. Turning them off can cause less smart players to rush in before you close off their escape and turn them back on.


Modèle:Jobs