Oldman Robustin wrote:Luke Cox wrote:Half the people who play AI will also bolt down all of security if anybody screams "harm" while they're being arrested. Asimov is ridiculously traitor-friendly so long as it doesn't catch you killing somebody.
That literally never happens.
There was an era when this was true, and as someone who played in that era I will still frequently appeal to the AI when I've been harmbatoned, shot by detective, beaten, etc.
The best result I've gotten since I returned is AI just repeating on radio, "Please do not harm prisoners" before promptly forgetting I exist.
Best luck I've had is convincing the AI that sec is going to execute you, but all that gets is the AI keeping your cell bolted in perma.
Locking down security can have severe consequences and achieves very little. Usually, or at least that is what AI players are forced to think, security has only the best interest at heart. Act in good faith and all that jazz. Law 1 does not allow you to pretend that you can peer into the future and determine that security will keep on harming prisoners.
If a prisoner is being killed in perma, you yell about it and maybe bolt down some airlocks. Again, that is all the power at your disposal. Set the harmful officer to arrest and hope the other officers deal with it. If they don't deal with it and order you to unbolt the brig, you have little choice but to follow it unless it is apparent that they are going to harm another human. And don't think a cyborg can just stroll in there and imprison the officer. It will die quickly, as well as the rest of your cyborgs shortly after. People are VERY quick on the 'detonate' button on the robotics console and I wouldn't be suprised if people had a "AI ROGUE" macro running.
Then there's the factor of obstructing security work, enabling possibly harmful antags do go on a rampage during your little, futile bitch fight with security.
If security is going full hitler mode, killing everyone, the situation is so far out of control that you cannot do anything anyway.
I consider myself to be a very neutral AI player and I rarely bolt down security - most of the time it's just not worth the risk and all the moaning from the officers. It does more harm than good. There's very antag friendly AI players too, that will even release perma prisoners on a vague notion of harm. And there's antag hunting AIs that will turn a blind eye as long as it ensures redtext.
As you can see the AI can swing either way, but in the end it's the officers doing the work and the AI has little to no influence over them.
It is easy for a security victim to blame the AI ('They harmed me and you didn't do anything about it!') as a last ditch effort, but the reality is that the AI, at best, is an inconvenience to the sec force.