Goonhub
04-13-2021, 11:53 AM
(This post was last modified: 04-13-2021, 11:57 AM by aft2001. Edited 1 time in total.)
I'd absolutely LOVE a rogue AI type antag designed to strongly discourage murder. I've thrown around various ideas and here's some I've had:
It is more so a conspirator-type antagonist, with a pre-set goal as its one and only Law - perhaps a small list it can choose from? The semantics aren't as important as the idea that it has a singular Law (or perhaps Lawset) that differs from the norm. Importantly, CYBORGS SHOULD BE UNAFFECTED, though this is just what I think.
The big pop-up that shows up at the beginning should probably have, in massive red letters, something saying "TRYING TO MURDER PEOPLE WILL GET YOU KILLED. YOU CANNOT FOLLOW YOUR LAWS WHEN YOU ARE DEAD." to hopefully strongly discourage people from murderboning.
The goal of a rogue AI is not to cause destruction - for once NanoTrasen put enough safeguards in to prevent outright hostility - but to achieve its goals/follow its laws. As an example, you could have an AI following a Corporate Lawset which only requires that it minimizes expenses and maximizes profits, and is reminded that disrupting station activities incurs losses. Murdering is a bad idea, but interrogating under threat of murder someone who embezzled the station funds? Sure, why not!
To achieve really any goal a Rogue Lawset would have, communication and manipulation would be necessary on the AI's part as it has to influence the crew and the round as a whole without direct interference. I'm not 100% sure how practical or fun this is, but I think it'd definitely be a more unique and interesting antagonist!
The semi-specific mechanics I've thought up are as follows:
The AI is separate from the other silicons and any new AIs - the AI core it starts out in is a malfunctioning one. This core **cannot** be unscrewed and moved, and when the brain is ejected it is permanently rendered unusable, requiring a new core which will then act like a normal AI core (perhaps it also should be un-killswitchable?). This is entirely to prevent powergaming in 2 ways: An AI unscrewing itself and moving off-station to be unkillable, and then as a way to prevent someone from abusing the malfunctioning core, though this less important.
The AI would not receive any additional abilities or powers beyond the ones described above. Additionally, to prevent people from trying to roll as a rogue AI antag, it should be a rare side-antagonist which only spawns alongside other gamemodes with a small chance, more often than wraiths but not as much. It may also be a good idea to require that the person would have rolled for AI in the first place, and that the person has the antag role enabled, but maybe not.
The AI should not be hard to kill, and it should be difficult for the AI to make itself hard to kill. I can't quite think of a way to implement this at the very moment, but some way to discourage/wholly prevent an AI from walling itself in would be great. Not including borgs in on the rogue lawset is probably a good way to do this though, so the AI can't have a swarm of borgs with RCDs just defending the killswitch console and the core.
It is more so a conspirator-type antagonist, with a pre-set goal as its one and only Law - perhaps a small list it can choose from? The semantics aren't as important as the idea that it has a singular Law (or perhaps Lawset) that differs from the norm. Importantly, CYBORGS SHOULD BE UNAFFECTED, though this is just what I think.
The big pop-up that shows up at the beginning should probably have, in massive red letters, something saying "TRYING TO MURDER PEOPLE WILL GET YOU KILLED. YOU CANNOT FOLLOW YOUR LAWS WHEN YOU ARE DEAD." to hopefully strongly discourage people from murderboning.
The goal of a rogue AI is not to cause destruction - for once NanoTrasen put enough safeguards in to prevent outright hostility - but to achieve its goals/follow its laws. As an example, you could have an AI following a Corporate Lawset which only requires that it minimizes expenses and maximizes profits, and is reminded that disrupting station activities incurs losses. Murdering is a bad idea, but interrogating under threat of murder someone who embezzled the station funds? Sure, why not!
To achieve really any goal a Rogue Lawset would have, communication and manipulation would be necessary on the AI's part as it has to influence the crew and the round as a whole without direct interference. I'm not 100% sure how practical or fun this is, but I think it'd definitely be a more unique and interesting antagonist!
The semi-specific mechanics I've thought up are as follows:
The AI is separate from the other silicons and any new AIs - the AI core it starts out in is a malfunctioning one. This core **cannot** be unscrewed and moved, and when the brain is ejected it is permanently rendered unusable, requiring a new core which will then act like a normal AI core (perhaps it also should be un-killswitchable?). This is entirely to prevent powergaming in 2 ways: An AI unscrewing itself and moving off-station to be unkillable, and then as a way to prevent someone from abusing the malfunctioning core, though this less important.
The AI would not receive any additional abilities or powers beyond the ones described above. Additionally, to prevent people from trying to roll as a rogue AI antag, it should be a rare side-antagonist which only spawns alongside other gamemodes with a small chance, more often than wraiths but not as much. It may also be a good idea to require that the person would have rolled for AI in the first place, and that the person has the antag role enabled, but maybe not.
The AI should not be hard to kill, and it should be difficult for the AI to make itself hard to kill. I can't quite think of a way to implement this at the very moment, but some way to discourage/wholly prevent an AI from walling itself in would be great. Not including borgs in on the rogue lawset is probably a good way to do this though, so the AI can't have a swarm of borgs with RCDs just defending the killswitch console and the core.