Proposed Change to Silicon Policy: The AI and Borgs are NOT Security

[TheBadPerson]

To start here’s 2 paragraphs from the AI wiki that I’m fairly certain most AI players have never read:

“As AI you have the power to strongly influence the round and you should always be aware of that and consider your actions before you ruin someone else’s fun just because it gives you that feeling of winning. Remember that the game is not about winning but about the RP and the experience of the round.

For example it can be a real killer if the AI calls out someone as the Traitor because it saw him doing something suspicious. If you see a traitor buying items at the start of the round, you can go ahead and just ignore it to be kind and allow them to have a fun round, as you don't get to be traitor everyday.”

I’ve clocked in over 1000 hours in game at this point, and I play a ton of antag and security. In my opinion the level to which the AI and Borgs are allowed to valid hunt currently is horribly balanced. I keep a loose mental tab of what round ends me as an antag, and somewhere between a third and half of my runs are ended soley due to AI intervention, with little security input. That is not ok, and it’s terrible gameplay. 
 

As it stands currently, the AI is 100% permitted to bolt in and call out any suspected antag it sees at any time, on any lawset. (Admins have confirmed this). There are zero rules against an AI scouring the station looking for suspected antags to bolt or call out at roundstart, and its laws even often support such behavior. 
 

While that may be seemingly realistic, it makes for terrible gameplay, and also has several issues. 
 

First off, AIs are not considered members of security, and get OOC protection from being preemptively killed or carded without an objective to do so. The problem is that they can round end any antag at any time with a well timed bolt or callout. This makes them effectively a silent AA detective that may or may not decide to end you at any time, and you’re not allowed to deal with them until they strike, which will almost always round end you regardless. 
 

This is a very unfair dynamic, as the antag has no chance to fight back against a random bolting or callout that can come at any time. You are effectively either forced to buy a syndicate teleporter, or hope the AI doesn’t feel like bolting or hunting. Again, you are not allowed to attack an AI preemptively despite it being allowed to take on security duties whenever it wants. 
 

Secondly, the dynamic of AIs hunting antags is terrible gameplay. The antag is silently and instantly cheated of their round by an untouchable eye that can completely lock them down, and security is cheated of an interesting chase or encounter. I have been on both sides of valid hunting AIs, and it’s boring for both security and the antag player. It is a dirty, cheap way to round end someone that is lame for everyone except the AI. 
 

What I propose for the sake of improving gameplay is this; add a clause to AI and Borg SoP that they are not allowed to interfere with security chases unless directly ordered to do so, and restrict callouts to wanted individuals. This will help restrict “detective” AIs, and prevent AIs from instantly round ending every antag they can find. I also think Borgs and AI should spawn in with bold red text that says YOU ARE NOT SECURITY similar to other roles as well. I would want this SoP reflected in all non-paladin/robocop lawsets. 
 

As it stands now this is what I see constantly: AIs will spawn in and IMMEDIATELY scour the station checking every single meta area for potential antags. They will see who asks for job changes at the hop and follow those people. They will look in offices and lower traffic areas for anyone with a PDA. They will essentially hunt for any excuse to sic sec on someone. They will leap at ANY sec callout and instantly bolt. The amount of AIs that play this way is disappointingly high, and in my eyes it’s basically rules-legal valid hunting since they can excuse it via lawset on technicalities, and it’s near impossible for admins to catch if they even consider it a problem at all. 

An AI or Borg should NEVER be the sole reason an antag goes down, yet it happens all the time. An AI or borg will take it upon themselves to catch an antag by themselves with zero security intervention or presence. I have seen this happen over and over and over. The AIs and Borgs that do this do not get bwoinked or even talked to. I have literally seen Borgs chase antags into maintenance to capture them on NT default by themselves and not get bwoinked. 
 

At the end of the day it I’m just tired of being round ended to valid hunting AIs, and I know many others are too. It’s not fun, it’s not sporting, and it’s really not fair either. Most people I ask about this topic express disdain at AIs as well for this reason. AI is the most powerful role on the station, and it should be played responsibly. It is currently not played responsibly by the majority of players in my opinion. There is a department dedicated to valid hunting, it’s called security, not the AI/Cyborg. I think the silicon rules should reflect that. 
 

Adding these miscellaneous statements here at the end since I suspect some very poor counter arguments to show up: 

-The ability to hack doors is not a counter to bolting. Bolting is used to cut off escape and allow sec to catch a fleeing antag. You will RARELY escape a proper bolting. 
 

-Yes this post is salty. No that does not invalidate the fact that getting round ended solely to AI is horrible gameplay. 
 

-Yes the AI can be emped, bombed, or carded. No that does not mean it isn’t still incredibly powerful, nor does it prevent it from instantly round ending you at the drop of a hat. Remember, the AI has OOC protection until it impedes you or you have an objective. 
 

-“Just hide your crimes or play stealthy!” The AI has overview of most of the station, and often actively hunts for anything remotely suspicious. You will be hard pressed to avoid the attention of a hunting AI.

-“One time the AI bolted me in with sec and I killed them all!” That’s great, but not every antag run gifts you with adrenals and chainsaw on discount. 
 

 

[Nerfection]

Without addressing your wider point, or touching on balance and things like that, I'm just going to respond to a couple of the rules/admins-based arguments:

Quote

As it stands currently, the AI is 100% permitted to bolt in and call out any suspected antag it sees at any time, on any lawset. (Admins have confirmed this)

I'm not sure which admin would have told you that, and my assumption is that this is at least a partial mistranslation of what admins have said. We fairly commonly bwoink bolt-happy AIs who do so too much, and/or without reason. I'd go as far to say that this is far and away the most common thing silicons get bwoinked for.

Quote

There are zero rules against an AI scouring the station looking for suspected antags to bolt or call out at roundstart

Other than Rule 8 which expressly forbids non-security to hunt antags, Rule 2 would also make using meta-knowledge to effectively powergame as an AI actively looking for antags which it would have no IC reason to believe are there bwoinkable. You are right, though, that this by itself is difficult to see or prove. However, an AI acting upon it would at least get a talking to. As one of the more major roles, AIs get a higher degree of scrutiny, so it's not at all uncommon for admins to do this.

In terms of actually stopping validhunting AIs - which I agree is a net positive for the server - there is one woefully underutilised method which can totally undo them: ahelping. If players actually ahelp AIs that go beyond their remit to just hunt antags then those AIs will either have to change their ways or not be allowed to play AI anymore. Admins aren't omniscient, and if we're not pointed to potential rulebreaks, they will often get missed.

Good luck with the thread!

Nerf

[TheBadPerson]

4 minutes ago, Nerfection said:

Without addressing your wider point, or touching on balance and things like that, I'm just going to respond to a couple of the rules/admins-based arguments:

I'm not sure which admin would have told you that, and my assumption is that this is at least a partial mistranslation of what admins have said. We fairly commonly bwoink bolt-happy AIs who do so too much, and/or without reason. I'd go as far to say that this is far and away the most common thing silicons get bwoinked for.

Other than Rule 8 which expressly forbids non-security to hunt antags, Rule 2 would also make using meta-knowledge to effectively powergame as an AI actively looking for antags which it would have no IC reason to believe are there bwoinkable. You are right, though, that this by itself is difficult to see or prove. However, an AI acting upon it would at least get a talking to. As one of the more major roles, AIs get a higher degree of scrutiny, so it's not at all uncommon for admins to do this.

In terms of actually stopping validhunting AIs - which I agree is a net positive for the server - there is one woefully underutilised method which can totally undo them: ahelping. If players actually ahelp AIs that go beyond their remit to just hunt antags then those AIs will either have to change their ways or not be allowed to play AI anymore. Admins aren't omniscient, and if we're not pointed to potential rulebreaks, they will often get missed.

Good luck with the thread!

Nerf

Wasn’t aware AIs didn’t get free reign to bolt and hunt at will. It was my understanding they could. I still think it’s a very grey area that needs to be addressed by SoP in some way to help reduce the behavior. I constantly see AIs aggressively hunting antags and taking them down solo, so it was my perception that they weren’t restricted to do so. Just last night an AI took down 2 nonviolent antags by itself. I ahelped it, no idea if anything came of it or not. 
 

My point remains that a large percentage of my rounds I find myself fighting the AI and Borgs more so than security. 

[Nerfection]

Quote

I ahelped it, no idea if anything came of it or not. 

I do think we as admins could sometimes do more to let players know what came of their ahelps. Perhaps knowing that the offending AI was warned/banned/or actually valid would have helped resolve that particular instance for you.

(I do agree that a 'You are not sec' text certainly couldn't hurt for silicons. After all, even Robocop only has following the law as it's third priority out of three)

[procdrone]

I still think adding an interraction timer on doors and equipment interactions (with malf upgrade to remove the timer) to be a great way to curb AI ability to fuck over antags.

[procdrone]

sorry for the doublepost.

The problem with the AI's is, both the rules and lawsets make it quite chaotic to understand on how you should play silicons. You're supposed to follow your laws, and most lawsets tell you to protect the crew and/or station, and since you're knowledgable on antags, you know they are going to hurt either. so the laws themselves tell you to stop these evildoers and that its something you're supposed to be doing. (even if its not fun for the receiving party)

On the other side of the fence you got the rules and admin rulings that outline how you should behave as silicon, you're expected to not tryhard too much but still follow your lawset. Its a very delicate balance to strike and most silicon players struggle with it. Which in effect practically makes you ignore whatever laws you have as flavours of practically the same behaviour sets.

Obviously, there are valid hunters that relish in the opportunity to win, but I feel like they are a margin of AI players, not the Majority.

I dont really see any clear way to improve silicon situation. It would require either heavy changes to the rules/wiki pages and giving clear directions for silicons, or some mechanical curbing to prevent being too OP. (PS, borgs are so easy in direct combat i dont think they are particulary risky enemy to deal with, AI would be the main problem here)

 

[Gatchapod]

I still maintain that the underlying issue is that AI is a rather boring job.
Cyborgs are self-sufficient. Crew expects you to set up T-comms once, open airlocks once every 10 minutes at most and do an odd request or two.
Bored AI is, whether you want it or not, pushed towards interacting with the most interesting and unpredictable part of the round - antagonists and security.

Ideal solution to this problem is not changing silicon policies, but rather giving AIs actual responsibilities to distract them from validhunting.
Sadly, this idea runs into many design problems the more you think about it.
AI tasks should be crucial for the well-being of the station at large to encourage focusing on them, yet not to the point of round-ending in case of an inexperienced AI. Additionally, crew should be able to fulfill these tasks in case of a missing AI. AI should ideally be able to simply complete them more easily.

As for the actual antagonists vs AI problem, it is one that many people struggle with in my experience. However, there are skilled players who plan ahead appropriately and stay a step ahead of the AI. It is possible to achieve, I have seen them do it multiple times. As such, I believe this to be largely a skill issue.
However, we could lower the skill floor here a little. Currently, the ones suffering from bolting the most are traitors. Most other antagonists can do relatively fine in my opinion. Due to that, we may want to consider a new uplink item - AI jammer. This would ideally outright prevent the AI from interacting with machinery around the traitor, letting people not worry about that much about being boltshocked the moment AI finds them trespassing.
Alternatively, procdrone's solution would be an even easier and simpler solution.

My point here is that I deeply believe AI problems can, with enough creativity and effort, be addressed via code solutions. Making more precedents and complex policies is a very artificial way of handling it and it is prone to failures. Players not knowing the policies, players not wanting to ahelp, admins not being available, to name a few.
If the code no longer supports certain actions, players have no choice but to comply with it.

Edited April 1, 2023 by Gatchapod

[TheBadPerson]

1 hour ago, Gatchapod said:

I still maintain that the underlying issue is that AI is a rather boring job.
Cyborgs are self-sufficient. Crew expects you to set up T-comms once, open airlocks once every 10 minutes at most and do an odd request or two.
Bored AI is, whether you want it or not, pushed towards interacting with the most interesting and unpredictable part of the round - antagonists and security.

Ideal solution to this problem is not changing silicon policies, but rather giving AIs actual responsibilities to distract them from validhunting.
Sadly, this idea runs into many design problems the more you think about it.
AI tasks should be crucial for the well-being of the station at large to encourage focusing on them, yet not to the point of round-ending in case of an inexperienced AI. Additionally, crew should be able to fulfill these tasks in case of a missing AI. AI should ideally be able to simply complete them more easily.

As for the actual antagonists vs AI problem, it is one that many people struggle with in my experience. However, there are skilled players who plan ahead appropriately and stay a step ahead of the AI. It is possible to achieve, I have seen them do it multiple times. As such, I believe this to be largely a skill issue.
However, we could lower the skill floor here a little. Currently, the ones suffering from bolting the most are traitors. Most other antagonists can do relatively fine in my opinion. Due to that, we may want to consider a new uplink item - AI jammer. This would ideally outright prevent the AI from interacting with machinery around the traitor, letting people not worry about that much about being boltshocked the moment AI finds them trespassing.
Alternatively, procdrone's solution would be an even easier and simpler solution.

My point here is that I deeply believe AI problems can, with enough creativity and effort, be addressed via code solutions. Making more precedents and complex policies is a very artificial way of handling it and it is prone to failures. Players not knowing the policies, players not wanting to ahelp, admins not being available, to name a few.
If the code no longer supports certain actions, players have no choice but to comply with it.

I am infinitely curious as to what you think the outplay is to an AI full bolting the captains office on an antique laser thief out of the blue is. 

People seem to love throwing out the term skill issue, and then have zero solution to said “skill issue”. Very interesting. 
 

I see skilled players go down solely to AI all the time. As I’ve said, it’s terrible gameplay for everyone except the AI. 

[Generaldonothing]

24 minutes ago, TheBadPerson said:

I am infinitely curious as to what you think the outplay is to an AI full bolting the captains office on an antique laser thief out of the blue is. 

People seem to love throwing out the term skill issue, and then have zero solution to said “skill issue”. Very interesting. 
 

I see skilled players go down solely to AI all the time. As I’ve said, it’s terrible gameplay for everyone except the AI. 

If you steal the laser that's the automatic alarm on display cases, not the AI

[Gatchapod]

32 minutes ago, TheBadPerson said:

I am infinitely curious as to what you think the outplay is to an AI full bolting the captains office on an antique laser thief out of the blue is.

Subvert the AI, cut cameras, cut AI control wires, jump into disposals, use thermite or C4, get creative.
AI relies on specific tools. You can deny it access to these tools or entirely bypass them.
If non-antagonists can help against antagonists, they absolutely should. If you plan to attack a crowded area, you should prepare to fight off multiple people. Consequently, if you plan to do something that might provoke the AI, you should prepare to fight off the AI. It seems extremely arbitrary to forbid AIs from interacting with antags they catch red-handed when everyone else is allowed to intervene.

Let me address something else, too. Current meta is to put AI on NT Default, because people expect NT Default to "do the optimal play" against antagonists. Sometimes Corporate for the same reason. If we add the suggested policy change, it will inevitably shift the meta in one of two directions:

1. Command/security issuing a blanket order to assist security however possible the moment antags are confirmed. Then, sometime later, another thread will be made in suggestions forum to somehow forbid that.
2. Command just defaulting to PALADIN/Robocop immediately to fight antags.

 

I believe the suggested policy change would be very quickly bypassed.
This is why I claim, once again, that only a code solution can actually affect the situation in a lasting positive way.

[Rufus]

As far as this topic goes I think there should at least be some large red text added at the start of the round for AI's and borgs that reads something along the line of

"Remember that you are not security, while you are able to assist when asked, do not willingly go hunting down EOCs or go bolting them in areas".

There have been far to many rounds where I have seen AI or valid hunting borgs single handedly ruin rounds for antags, and while some of it comes down to the skill of the player like with how Gatchapod said, not every situation is escapable and 90% of players will not have the skill to do so. A recent example that comes to mind for me was when I was being chased by a borg who flashes me so I cant see, while running from five to six sec officers, then while trying to escape down the hall the AI bolts every door, leaving me at a dead end with the only way out now being to go through all of security. This is not fun gameplay to have to deal with.

There have also been far too many borgs that I see while spectating going around valid hunting mid rounds, such as a medical borg attacking a bio hazard nest with no crew inside, alone in toxins. Or another example being an engineering borg welder bombing a xeno nest before the bio hazard alarm was even sounded, borgs chasing down antags to flash spam them, etc. While alot of this comes down to the player I feel that valid hunting silicon's are way more common than normal crew because of possible "crew harm", and having to deal with valid hunters that don't slow down when they are hurt, can bolt and shock doors to prevent escape, and in the case of the ai only being able to be delt with IC by attacking highly secure and defended areas far out of your way normally.

end of the day I just think something should be done because bottom line, its not fun gameplay.

Edited April 1, 2023 by Rufus

[TheBadPerson]

4 hours ago, Gatchapod said:

Subvert the AI, cut cameras, cut AI control wires, jump into disposals, use thermite or C4, get creative.
AI relies on specific tools. You can deny it access to these tools or entirely bypass them.
If non-antagonists can help against antagonists, they absolutely should. If you plan to attack a crowded area, you should prepare to fight off multiple people. Consequently, if you plan to do something that might provoke the AI, you should prepare to fight off the AI. It seems extremely arbitrary to forbid AIs from interacting with antags they catch red-handed when everyone else is allowed to intervene.

Let me address something else, too. Current meta is to put AI on NT Default, because people expect NT Default to "do the optimal play" against antagonists. Sometimes Corporate for the same reason. If we add the suggested policy change, it will inevitably shift the meta in one of two directions:

1. Command/security issuing a blanket order to assist security however possible the moment antags are confirmed. Then, sometime later, another thread will be made in suggestions forum to somehow forbid that.
2. Command just defaulting to PALADIN/Robocop immediately to fight antags.

 

I believe the suggested policy change would be very quickly bypassed.
This is why I claim, once again, that only a code solution can actually affect the situation in a lasting positive way.

Adding tasks via code is not going to keep AIs from valid hunting. The sole reason a lot of people queue up AI is to bolt and catch antags. You can add all the tasks you want to AI, they’re just going to ignore them to go valid hunt. 
 

I want actual rules in place that establish that an AI is not supposed to be attacking antags. I don’t even think paladin and robocop should exist. Synthetics completely break the game balance between security and antags, and I’m tired of having to deal with the toxic gameplay of being hunted by an AI. 
 

The counters you listed are also not counters. If you attempt to do any of that against a bolting AI you will get roflstomped so fast your head will spin. 

[TheBadPerson]

4 hours ago, Generaldonothing said:

If you steal the laser that's the automatic alarm on display cases, not the AI

Other day I saw a stealth antag steal the laser and Jetpack. AI immediately bolts the entire office down screwing the antag, who is promptly firing squaded by sec. That’s what I mean by this. I don’t see why that’s ok, and it’s terrible gameplay. 
 

That dude literally got round ended solely to a valid hunting AI, and that sort of thing happens CONSTANTLY. 

[Gatchapod]

10 hours ago, TheBadPerson said:

Adding tasks via code is not going to keep AIs from valid hunting. The sole reason a lot of people queue up AI is to bolt and catch antags. You can add all the tasks you want to AI, they’re just going to ignore them to go valid hunt. 
 

I want actual rules in place that establish that an AI is not supposed to be attacking antags. I don’t even think paladin and robocop should exist. Synthetics completely break the game balance between security and antags, and I’m tired of having to deal with the toxic gameplay of being hunted by an AI. 
 

The counters you listed are also not counters. If you attempt to do any of that against a bolting AI you will get roflstomped so fast your head will spin. 

I believe you're suffering from a confirmation bias here.
Yes, some people play silicons to fight antags. As Nerfection said, this is not fine. Ahelp them and administrative actions will be taken.
However, I still maintain that more often this is an outcome of an otherwise boring gameplay. Furthermore, adding actual tasks for AI would already support taking additional administrative actions against them. If AI is hunting antags at the expense of doing its job, that is a clear rule break.

I'm sorry, but slapping more OOC padding on antagonists will not improve the gameplay. OOC padding inevitably breeds even more resentment.
"Toxic gameplay", as you say, is a result of your own vitriol. You refuse to learn, improve and adapt. You want admins to save your greentext. It's quite literally a buzzword (phrase?).
Even if we agree, hypothetically, that gameplay is toxic, this only further reinforces necessity for mechanical solutions to the problem. Policies could fix toxic behaviours. Toxic gameplay requires different solutions, since the issue clearly lies in how the game plays.

You're also plainly wrong about my suggestions not being counters. AI has no way to counteract sabotage of its equipment outside of having an engineering cyborg or hoping crew engineers help it.

 

I cannot believe you're making me defend silicons, you know? My street reputation of a sillycon hater is ruined. Anyway, I wrote my rants and don't plan on inflating the thread any further, just in case someone actually wants to read through it.
Good luck, even if I largely disagree with how you are trying to fix the problem.

[Hrulj]

This is Medium RP server. From how I see it Crew and AI should report suspicious behavior. If you saw your coworker pulling a gun or fiddling with a bomb in company bathroom you'd not think about the fun they'd have but the danger they pose. As it is crew already generally ignores antags or actively aids them to a great degree and hassle security for no reason even when station is in mortal danger ranging from simple blocking to slipping, wetting the floors or being a nuisance and I've even seen crew spawn monkeys and throw them at xenomorphs as a ghost or voluntarily feeding a vampire. In the hours I played I have seen AI actively hunt antags twice. Not counting antags breaking into AI sat or upload. Playing as sec I have never gotten a single notice by AI of suspicious behavior, despite traitors commonly using PDA's and AI being able to raid those. 

 

It would really impact my RP negatively that AI is just allowing antags free reign of the station despite the bloodshed they are doing. A vampire playing sec for suckers should get shocked by AI. Don't want it? Don't be obvious and get some insuls. 

Edited April 8, 2023 by Hrulj

[Sirryan2002]

What exactly else is the AI supposed to do? Their entire job is designed around watching crew members, specifically crew members who are causing issues on station, antag or not. Borgs can very easily just go do a task like mopping floors or repairing the station but the AI is locked in its observer role the entire round. So the most common way to solve this is by involving themselves most actively in whatever exciting thing is going on in the round, which is usually antagonists.

I don't disagree that AIs round-ending antags can be exhausting and make for poor gameplay, but additional rules are not going to fix this. The reason being that it would be a bad rule as it works directly against human nature in attempt to correct what is a game design issue at its foundation. AI, like most things in this game, needs to be overhauled.

Furthermore, this is a nightmare to enforce, I am not log-diving to see if Security Officer X gave the AI specific permission to intervene; nor am I going to check if the HoS just gave a ultimatum for the entire round for the AI to intervene whenever they feel. This is one of those situations where the effort needed to properly investigate rule breaks of this sort far outweighs the actual value of enforcement. This is not going to change the current AI -> Antag gameplay culture in the long-term as it will need to be applied to each and every AI player since it's not going to be a rule one just easily understands at first.

I actually think this will have the opposite effect of making AI players more toxic as they begin to target security players for not giving them the go ahead to intervene and "ruining" their AI round by neutering them of everything fun. If players are as you describe, this is exactly what is going to happen.

[MattTheFicus]

If you notice an AI is bolting your fellow Antagonists, simply go kill the AI. Just like you can kill Security Officers/Validhunters as an Antagonist, the AI can be punished for sticking its nose into your crimes.

Enough Iron + Uranium and/or ahelping about using explosives and AIs quickly stop messing with you. That, or they get carded and wiped/lathe'd.

[TheBadPerson]

9 hours ago, Hrulj said:

This is Medium RP server. From how I see it Crew and AI should report suspicious behavior. If you saw your coworker pulling a gun or fiddling with a bomb in company bathroom you'd not think about the fun they'd have but the danger they pose. As it is crew already generally ignores antags or actively aids them to a great degree and hassle security for no reason even when station is in mortal danger ranging from simple blocking to slipping, wetting the floors or being a nuisance and I've even seen crew spawn monkeys and throw them at xenomorphs as a ghost or voluntarily feeding a vampire. In the hours I played I have seen AI actively hunt antags twice. Not counting antags breaking into AI sat or upload. Playing as sec I have never gotten a single notice by AI of suspicious behavior, despite traitors commonly using PDA's and AI being able to raid those. 

 

It would really impact my RP negatively that AI is just allowing antags free reign of the station despite the bloodshed they are doing. A vampire playing sec for suckers should get shocked by AI. Don't want it? Don't be obvious and get some insuls. 

AIs hunt antags all the time, and crew are FAR more likely to try and valid an antag than provide any help whatsoever. I question how much you play sec or antag to have the opposite opinion. 
 

RP has nothing to do with the discussion. There are a TON of things that happen on server that fly in the face of RP for the sake of gameplay over realism. 

[TheBadPerson]

2 hours ago, MattTheFicus said:

If you notice an AI is bolting your fellow Antagonists, simply go kill the AI. Just like you can kill Security Officers/Validhunters as an Antagonist, the AI can be punished for sticking its nose into your crimes.

Enough Iron + Uranium and/or ahelping about using explosives and AIs quickly stop messing with you. That, or they get carded and wiped/lathe'd.

Theoretically correct. Haven’t seen someone do this successfully in a VERY long time. Just because something is possible doesn’t make it realistic or a solution. If an AI messes with you it will almost always round end you if the AI has any remote idea what they are doing. 
 

I maintain that AI interference is awful gameplay. Nobody I know who antags likes it. It would be one thing if they did it to stop murderboning, but that’s never the case. 

[TheBadPerson]

7 hours ago, Sirryan2002 said:

What exactly else is the AI supposed to do? Their entire job is designed around watching crew members, specifically crew members who are causing issues on station, antag or not. Borgs can very easily just go do a task like mopping floors or repairing the station but the AI is locked in its observer role the entire round. So the most common way to solve this is by involving themselves most actively in whatever exciting thing is going on in the round, which is usually antagonists.

I don't disagree that AIs round-ending antags can be exhausting and make for poor gameplay, but additional rules are not going to fix this. The reason being that it would be a bad rule as it works directly against human nature in attempt to correct what is a game design issue at its foundation. AI, like most things in this game, needs to be overhauled.

Furthermore, this is a nightmare to enforce, I am not log-diving to see if Security Officer X gave the AI specific permission to intervene; nor am I going to check if the HoS just gave a ultimatum for the entire round for the AI to intervene whenever they feel. This is one of those situations where the effort needed to properly investigate rule breaks of this sort far outweighs the actual value of enforcement. This is not going to change the current AI -> Antag gameplay culture in the long-term as it will need to be applied to each and every AI player since it's not going to be a rule one just easily understands at first.

I actually think this will have the opposite effect of making AI players more toxic as they begin to target security players for not giving them the go ahead to intervene and "ruining" their AI round by neutering them of everything fun. If players are as you describe, this is exactly what is going to happen.

Then I don’t know the solution. What I do know is getting round ended solely to an AI bolting your escape, or an AI being sole witness to an otherwise silent crime is just awful gameplay. This happens WAY more often than people like to admit. 
 

And despite what people like to convince themselves, there are not many viable counters against an aggressive AI outside the syndicate teleporter, or vampiric teleports/walks. 
 

I am not the best player on the server, but I’m pretty damn competent. People loveeeee to throw out “skill issue” when you complain about the AI, yet they themselves get annihilated when the same happens to them. They have no solutions. The amount of people I have seen survive an aggressive AI is LOW, and it’s always been through luck, or extensive chaos. 
 

At the end of the day I hate that I always have to accept that an AI can round end me whenever it feels like the second it decides it wants to CTRL-Click. I am not the only one that feels that way. Basically every other player I know of that antags frequently DESPISES the AI. 

[Hrulj]

50 minutes ago, TheBadPerson said:

AIs hunt antags all the time, and crew are FAR more likely to try and valid an antag than provide any help whatsoever. I question how much you play sec or antag to have the opposite opinion. 
 

RP has nothing to do with the discussion. There are a TON of things that happen on server that fly in the face of RP for the sake of gameplay over realism. 

That is factually incorrect. Just go read ban appeals section. Valid hunting is the easiest way to get pinged. Up until a few weeks ago I exclusively played sec and my experience has always been total shit when it comes to the way crew treats you and the antags. Even people murdered by antags and revived literally don’t say anything or report a thing about who murdered them. They just get cloned and get back to working or sitting in front of bridge. Be an observer and ghost 5 rounds with ghost huds on. Tell me crew ain’t helping antags way more than reporting them after observing that. 
 

AI has like 4 sets of laws that would allow it to ignore actions of antagonists. Everything else be it corporate, crewsimov, Paladin or whatever sets first or second priority as dealing with threats to station or profits. 

[CodeLyoko]

As someone who has played a lot of silicons as a borg, I will say in my experience the AI is normally telling the borgs to avoid antags. be it a crazy chainsaw man or a emagger. I have on multiple occasions been told to leave maints as a engineering borg when looking for cut wires. I have seen plenty of AI's threaten to lock valid hunting borgs for violating the "keep yourself safe" law that most lawsets have. I've only ever been told to "hunt antags" once on a non-robocop/paladin lawset and refused due to it breaking the "keep yourself safe law".

 

As a borg, I have been bwoinked myself for fighting an antag by an admin, so it is not like the admins are not watching for borgs vaildhunting, as I have experienced getting a talking to about it before. I don't think adding more rules where the AI cant report suspicious people will solve anything other then making AI an even more boring job, or as others have said, just have the AI neuter security for not letting them report bad people.

[Veterankyl]

I think one of the biggest issues is deciding at what point SHOULD an AI get involved if not directly told to. If it's minor crimes or something that security can handle without any help, then an AI should probably at most just alert security, just like your average bystander would. But what happens when an antag starts doing things that go against AI laws, or an antag becomes so powerful that security can't effectively fight it on their own? If an antag has killed several officers, shouldn't the AI bolt them in so they can't go do further harm to the crew or station? Would only alerting security be considered inaction if the AI has more tools at their disposal to help? I think the AI should be allowed to involved themselves, but only when an antag is clearly too strong to be handled by security and is doing things that go against the AI's lawset.

[TheBadPerson]

20 minutes ago, Veterankyl said:

I think one of the biggest issues is deciding at what point SHOULD an AI get involved if not directly told to. If it's minor crimes or something that security can handle without any help, then an AI should probably at most just alert security, just like your average bystander would. But what happens when an antag starts doing things that go against AI laws, or an antag becomes so powerful that security can't effectively fight it on their own? If an antag has killed several officers, shouldn't the AI bolt them in so they can't go do further harm to the crew or station? Would only alerting security be considered inaction if the AI has more tools at their disposal to help? I think the AI should be allowed to involved themselves, but only when an antag is clearly too strong to be handled by security and is doing things that go against the AI's lawset.

Experienced AI players play this way, and only start to involve themselves when it is clear the antag is getting overly violent, or sec clearly cannot handle the issue themselves. This is appropriate to me. The issue is this is purely an honor rule/cultural thing done out of respect for other players and is completely optional. 
 

What I see far more often is AI immediately resorting to bolting/spying without really any reason to do so. Believe me when I say that security RARELY needs help outside their standard armory options. AI intervention is almost always massive overkill. 

[sirius]

17 hours ago, TheBadPerson said:

Experienced AI players play this way, and only start to involve themselves when it is clear the antag is getting overly violent, or sec clearly cannot handle the issue themselves. This is appropriate to me. The issue is this is purely an honor rule/cultural thing done out of respect for other players and is completely optional. 
 

What I see far more often is AI immediately resorting to bolting/spying without really any reason to do so. Believe me when I say that security RARELY needs help outside their standard armory options. AI intervention is almost always massive overkill. 

AI players me included often do something like that (even if they dont tell sec immediatly cause that would be boring for both OOC) because they are extremely bored cause AI gameplay boils down to spying on people. AI has 0 goals, except when i decide to, for example, micromanage engineering cause there is no CE or something similar. AI players just exist in a 2 hour boring vacuum of spectating and opening a door once every 20 minutes so its no wonder they focus on the only interesting thing happening: the antags. IMO AI should have some minigame or seperate thing to do to unlock some tech that malf AIs have like repairable cameras and such so that an AI player isnt just existing waiting for a door command. Once the AI is actually needed in can focus back to the actual happening of the station but while everything is fine it does its seperate thing. How that "minigame" or whatever should look like i have no idea but giving AI players something seperate to do would solve the problem of the validhunting AI