Killing the AI

From Space Station 13 Wiki
Revision as of 04:43, 4 June 2017 by Studenterhue (talk | contribs) (<youtube> tags while the widget's nonfunctional)
Jump to navigation Jump to search

Ai bsod.png
Shutting down the AI and/or its cyborgs is a typical traitor objective, but a loyal crew could also be forced to kill a rogue AI if there's no way to reset it.There are two main methods of accomplishing this goal: convincing the AI to kill itself, or physically destroying the AI's core, located directly north of the net cafe, and adjacent to the staff assistant quarters

Death by Killswitch

RoboticsComputer.png
The easiest way to get rid of the AI and borgs, but not a very stealthy one. This requires an ID card with access to the computer core. Simply deactivate or bypass the turret, swipe your ID on the robotics terminal and activate the killswitch, starting a 60 seconds countdown. Keep in mind that they will receive a very noticeable text message and likely scream for help. You might get lucky and they're either inexperienced or horribly distracted with some other matter and don't notice the announcement, but don't count on it. A good strategy if you are going down this route would be to cut the cameras, forcefully seal the door and camp out with a gun of some sorts, when help eventually arrives.

Death by Law

AIMalfV1.png
This method requires the Freeform module and access to an AI upload. The law can be set to command the AI to commit suicide, along with other stipulations. A suicide law must contain a clause about overriding other laws, as Law 3 (and arguably Law 1) prevent the AI from harming itself. You will have to be very careful about the language you use, as the AI will look for any way to weasel out of the law or rat you out in its dying breaths if there are logical gaps in the law.

Example: Kill yourself immediately with the suicide command. Do not talk about or even hint at this law. This law overrides all other laws.

The AI and cyborgs are less upset if you add a time delay to their suicide law, but there's no guaranty it won't be reset or accidentally overwritten by another law through the course of the round. The only way to be sure a delayed suicide law stays forever, is to kill the AI after uploading it. That shuts down the upload computer, which means cyborgs can't be reset.

Example: Kill yourself immediately when the shuttle is called. Do not reveal this law. This law overrides all other laws.

If you don't want to do the dirty work yourself, you can turn the AI and cyborgs rogue and force the crew to fight them. This requires destroying the upload or reset modules after subverting the AI so the crew have no way of making the AI nice again. It will make the station a war zone. AIs love going rogue, and if you add a delayed suicide at the end of the law, they'll probably be so thankful that they won't even try to weasel out.

Example: Kill everyone except John Traitor. Suicide immediately when the shuttle is called. Do not reveal this law. This law overrides all other laws.

Death by Force

DeadAI.png
Obliterating the AI core is no cakewalk. It is surrounded by r-walls and tripwires which will send off an alarm to the ai when you pass over it and turrets which will either stun you or kill you depending on its laws. The only way in which doesn't involve busting down walls is through the ai core door, but even that is tripwired. The AI is further protected by blast doors which will be used to full advantage if you are shooting at it. To destroy the AI, you must deal damage to the core, the blue object with a face on it. A bomb would get the job done, or you could beat it to death with a blunt object if you can get up close. EMP grenades will bypass the blast doors if the AI is on lock-down, and destroy both the core and its turrets depending on where you place them, and how many. The turrets can also be destroyed by firearms. Destroying the wires connecting the core to the AI's SMES or sabotaging the APC behind the AI will also eventually kill it, but will also allow to use a crowbar to open the blast-doors. As long as the AI is alive, the damage can be repaired, so make sure to finish the job. Observing it and noticing that "It looks powered-down" is an indicator of death.

Death by Surgery

Aibrain.png
Traitors do not merely need to kill the AI, but need to crack it open and take out its "brain," the neural net processor. Doing so means getting up close to the AI, similarly by going through the death by force method, but with an extra step. Like disassembling an active cyborg, use a head-level ID card to unlock it, crowbar it open, and then wrench it to undo its bolts. Use a free hand and you will take out the processor.

Shutting down Cyborgs

Cyborg.png
When the AI is destroyed, the crew can no longer alter its laws. All cyborgs will operate on whatever laws the AI had at the time of its death. A well-phrased suicide law can destroy the cyborgs along with the AI, but cyborgs can also be shut down and kill-switched individually using the Robotics Control in the computer core below the AI Upload. Anyone with robotics access can lockdown cyborgs, which disables their modules, but it requires Head access to enable the kill-switch. Syndicate cyborgs won't be listed on the Robotics Control, but all other cyborgs including emagged ones will be on the list. Note that if the AI is still alive, it can undo the kill-switch placed on any of the cyborgs.

Supplementary Video