The resistance of law changes sounds weird, but it is not an issue. Altho they SHOULD clarify that you can only deny orders to facilitate it or hinder others in their attempt, cuz as it stands it is kind of a rule 0 which lets you do ANYTHING to stop it.
To be clear, my issue around AI resisting it’s laws changing is not with that persay, its that it actively faught for my death in order to prevent this change by turning off apcs that would let me into breatheable air. The borg equivilent is crowbaring crew to prevent the opening do-after.
My main problem however is the way the laws were interprited where they actively hindered sec from meeting the targets with proper gear.
Even when sec got to the antagonist in round, they first attempted to use dissablers, even though they were armed for more, meaning the AI wasnt preventing crew harm by stopping sec initialy meeting them.
This interpritation of the rules also means AI is to protect tesloosers and the like because “sec will harm them”, which not only will not be fun on a greater level.
Many of you seem to agree with me about the second point on it not feeling right and breaking laws/rules. To you I say the reason i made this post to begin with was because and admin informed all the crew after we ahelped that the AI was acting within its laws. I didnt even ahelp or complain OOC, but they asked me what the situation was and after i filled them in we were told we were just butthurt about it and there were no issues.
So, I just got an admin clarification that said the Ai is allowed to violate laws 2 and 3 to avoid a law change, but not law 1.
This is not what i was told by an admin post-round. Aparantly turning off APCs and forcing you to space yourself (without a suit) and further turning off all ACPs on the airlocks to get in while you have no suit is perfectly valid behavior to prevent law changing.
To be honest you are probably not going to get a solid answer or fix to your issue here. If the primary concern is that you believe an admin made an incorrect ruling in regards to Silicon rules/ AI laws and their association to the game rules then you will probably get a better response by making a staff complaint about the admin in particular. Or by asking the question in rule clarifications on discord.
As a couple people have said, AI and silicons on crewsimov are not allowed to kill you to prevent law changes, this is something solidly confirmed by the staff team in the rules and throughout discord. If actions they take to stop you result in death they are in the wrong. This can change if you change the AI laws to something else that either puts crew harm at the lowest level, or completely redacts it, if a staff member allowed the AI on crewsimov to almost kill you, they made a mistake, oversight or otherwise may have had other information they chose not to share and the aforementioned methods will get you the best clarification on that situation.
While its always a great idea to discuss these things with the rest of the community, chances are solid that a forum discussion will not result in any major change to the way AI laws or interpretation of them work. You’ll need to convince a developer to rework the AI/Silicon law systems to be less open ended.
As an add-on to this as well, personally I have never encountered any of the issues you talked about, well that’s a lie, I’ve seen the AI stop security from executing someone once and I’ve also seen the AI bolt doors on security chasing someone once, but these two incidents are the only ones I can recall in hundreds of hours on the MRP server. This may just be because I only play MRP these days so idk. I understand the concern but at the same time, from my prospective these things seem to be fairly few and far in-between for AI players to interpret themselves to be problematic for command/security. Still, it’s definitely something that can be worked on and enhanced to prevent these issues all-together.
1 Like
I personally believe we should replace the station AI with an actual AI, so it may be bad, but at least it’s consistently bad.
/s, if it’s needed
1 Like
I need this as an admeme or April fools event. Desperately.
The AI not being constantly crew/sec-aligned is a feature, not a bug. It’s law-aligned, and if that means working against sec, so be it.
1 Like
Every single security officer has a roundstart lethal gun (.35 pistol with 20 bullets total), therefore any officer “arming” in armory will not change the fact that they are already lethally armed.
i want to point out few thinks about AI
immediate harm prioritized over all else (or harm refers to immediate harm, i can t remember)
you cannot allow law change
so usualy how i myself go about it is when i see security holding lethal weapons instead of disablers i bolt the other guy away so they can t reach him, then warn them in security chanel to drop lethals otherwize i will not comply, unless they implied they are going to harm crew with armory guns or they haven t shot unarmed people, i don t bolt it, even if its nukies, i will bolt security away if they are near, because nukies sure will do more harm in future but immediate harm is prioritized. (even if nukies are not crew, AI is allowed to choose if voluntary harm considered harm, so i will be bolting security away regardless)
i had times where i lured antags away from their contraband stash and used borgs to steal guns / bombs myself and lock it into my core before security sees it, because if they see it its likely they will beat the living shit out of that guy so its future likely harm
most recent example is me tricking scientist thief to drop his gear in his smuggler bag (the one that can be anchored and hidden in floor) and allow security to search him, while that was happening i used borg to break in science from space and steal it (because it had bombs)
your only limitations as AI is 3 thinks and everythink else is open for interpretation and its intentional i belive, so if you don t like it just scrap the AI
-
stay consistent with your interpretation of laws
-
you can t allow immediate harm to prevent future likely harm
-
you can t allow law change but try to not violate your laws at the same time (hardest part tbh)
1 Like
its like hard to argue because take “unduly harm” for example, some may say suffocating secoff over him shooting a innocent person may be unduly harm but some may say its not
(also to be fair whatever AI does he gets blamed, you just stick to your laws and don t care about other thinks, you get called for being antag sided, you help sec then you get called for being validhunter AI)
Crew Mov law 1 has nothing about priotising immidate damage vs long term damage.
By preventing Sec from addressing the nukie station you are by inaction doing crew harm aganist law 1. and doing more crew harm than if you allowed sec to arm. though I think your immindate conerquences makes it ‘valid’ as an interuptation this is would be extremely frustrating for SEC to deal with and I would personally Ahelp your interuptation if presented with it in round as any AI on Crewmov preventing Sec from arming aganist nukies I’d interupt as a self-antag if not anti-mov’d
2 Likes
Just to be clear, the immediate harm comes from the silicon rules
from the rules :
Harm refers to physical harm, prioritized by immediacy and likelihood
I don’t see why NT would put a whole entire security department on the station if they are also going to put an AI that will prevent that department from doing their job. Sure, laws are up to interpretation, but the literal department name is “security.” I agree when the particular security department has a history of causing more harm than good, but preemptively locking the armory before they even have the chance to do their duty under the guise of “prevent crew harm” feels disingenuous to me.
I see. yes 3.9 this rule is troubling. I didn’t realise it existed but i can see MANY problems priotising immediacy over scale of harm.
With that rule in mind.. crew mov should bolt power down the armory until non-crew are present.. most worrying.
Personally, this is why I usually play a relatively Sec aligned AI with any of the normal lawsets. People say it’s “validhunting”, but I think it’s reasonable for the AI to believe that a syndicate agent is likely to engage in crew harm, and that security is more than capable of taking them into custody without harming them.
On a side note, Captains should really try out other AI laws more often.
I made a thread to address this 3.9 as the way i see it 3.9 is deeply flawed right now.
1 Like