[1 dismissal] A.I. Whitelist - Revisited?

[Jupiter Storm]

I realize this has probably been suggested before and failed, but I have several good reasons for revisiting the topic, due to the current environment, if I may?

The Context:

Partly this is based on certain rumours that have been thrown around regarding an A.I. rework (or removal) essentially placing them as a 'constructable' module in Robotics. I am assured that these rumours are just, exactly that - however one of the points that was made during one example of said rumour was really valid, so it almost made sense that it was going to happen:

Spoiler

All of these are really good points, though regarding the A.I.'s utility is up for debate.

The point is, these rumours became so prevalent that I had begun to take steps in modifying M.I.St.R.E.S.S.'s own suite of available functions, altogether refusing to note trespasses, observe the activities of non-crew and refusing to track people for any reason other than medical emergencies. Along with this, I was logging how badly this disrupted the round and copying the relevant Game ID's to post here, as evidence backing a suggestion at a later date. Furthermore, I have also been logging how much the activities of other departments are inhibited by the lack of A.I. involvement (heavily, so far) - I am continuing to do this, so that if anyone asks for the research I have been performing I can provide evidence later... But it's no longer strictly relevant.

So, to the point at hand!

As I said, all of the points that were made regarding A.I. are very good ones, but the salient issue seems to be, to me, with the propensity of players to powergame and shut down antagonist gimmicks almost instantly. Not to mention how horrendous some Malf rounds can be - and when you see people mass cryo'ing as soon as it becomes clear the A.I. is rogue, there's a serious issue which cannot be ignored.

I have done it myself - I have been horrified at how completely I've destroyed gimmicks in the past, and every time it's happened I draw myself down a little bit, or take a break and re-evaluate how I play the A.I. One round springs to mind in particular, where a Raider team breached the Command Docks and set off an air alarm, and I reported this - everything was locked down and the team just shot each other after arguing. It was so upsetting to see and I felt genuine remorse, both ICly and OOCly.

Why Whitelist the A.I.?

Firstly, I'm sure the issue has been discussed already by staff regarding the current issues caused by A.I. who push that little bit too hard. Particularly of late, where a large number of malf rounds have gone absolutely horribly and haven't been fun for anybody. Even if the idea was rescinded before, it seems a good time to ask the rest of the community what they think again - given that these rumours of reworking it are spreading, untrue as they may be.

Secondly, as was stated, the issue is with the players handling the A.I. Yes, it's a very powerful role, yes, it has the ability to entirely alter the dynamic of a round. This is why Command is whitelisted - due to the effect that their decisions have upon a round... Right? Given the ways in which the A.I.'s abilities (even unmalfed) can be used to restrict people heavily, you want somebody responsible and mature in the driving seat.

... And yes, I'm well aware that there are some people who will say, that rules you out, Jupiter, because of the kind of A.I. I play. The idea still stands, and I'm willing to be judged the same way as everyone else.

In Summary, doing what the rumours 'suggest' doing would not actually remove any of these issues, they'd merely delay the activation time of an A.I. Frankly this would just be frustrating for people who are more passionate about A.I. and like to actively roleplay and try to improve a round using their vast knowledgebase and abilities.

If you Whitelist it - you can ensure you get people who are vetted and experienced with Laws, Station Systems, Communicating with others well, etc. If they fall out of line? Remove it, they can't play. Make them apply again, or blacklist them more harshly!

I want to see a healthy move in the right direction to change A.I. from a Low RP tool designed to hammerblow antagonists and crew alike, into another kind of character, which I am sure many Synthlore Developers would be passionate about seeing as well. There are so many things that an A.I. can do, if played by someone passionate, who likes to facilitate roleplay and learning for others, beyond simply opening doors and bolting them shut.

Edit: One point I forgot to note - which is actually central to this entire argument - the suggestion of Whitelisting the A.I. places power over who gets it into the hands of the Synthlore Developers who are interested in promoting quality synthetic roleplay and characters that enrich rounds, as WELL as moderators who will assess the maturity of person and their capabilities in self-restraint.

Edited August 29, 2019 by Jupiter Storm

[Carver]

Once upon a time, roughly about 4-5 years ago, I considered the AI to be detrimental for the balance of things. Now that everything else station-related has been nerfed to shit, I fear how helpless the crew truly would be with less (or in the case of rumours, no) AIs. They're one of the few credible threats left to antagonists to keep them in check, and I stand by my belief that the game should not be centered around the gimmicks and whimsies of antagonist players (unless they themselves were whitelisted).

Is the AI balanced? Hardly. But it's there to support the crumbling framework around it of heavily nerfed non-antagonist tools. Do I think AI should be whitelisted? No. I understand why it would be wanted, but I'd rather not see it relegated to a small pool of players, and I'm of the belief that you can only really get away with having either Command or the AI whitelisted, and not both. It's not uncommon in lowpop that there's no Captain or HoP and only an AI. Now, the moment that the station departments (Security, Medical and Engineering) receive a credible anti-antagonist buff (whether greater access or better tools), is the moment I may support a whitelist (or the rumour suggestion).

[Guest Marlon Phoenix]

I agree with Carver.

[Naelynn]

Hello,

I believe this is a topic I can have a relevant input on - as a frequent antag player. 

Tl;dr:
DO NOT REMOVE OR WHITELIST THE AI.
PLEASE.
Sincerely: Antagonist player

Now for more details:
Ai's are vital for functioning of many departments and have a whole suite of aux functions that many people seem to ignore because a good Ai does them so well you won't even think about it. Ex: Pumps in atmos, mining airlock, configuring cryo cells in medical, assisting in locating blobs, rescue operations etc. etc. Seriously, play an ai and try to keep up with the ridiculous amount of things an entire crew worth of people will want from you - it's probably the closest you can get to an admin handling tickets alone on high pop.

Furthermore: Ai is an excellent role to LEARN COMMAND. It is the only non-command role to have access to their channel, and can observe how things are to be handled. This is suuuuper important for people who are looking to whitelist for command. You can say that Warden is the 'Learn Command' role, but... not really, even though they have access to fax machines.

As for shutting down antags:
Can Ai's do it? Absolutely. Is the end of the world for an antag team? No. 
MISTRESS in my yesterday's round as Vox Nightingale [raider] was in this 'maintenance' mode and it felt /Worse/ to play for me. 

Normally when Mistress is in a round and I'm antagging, I do have to really worry about her, because she has absolutely no qualms with wrecking my plans and telling security where to run, which usually after an hour-or-so of the round leads me decide that ai has to die [And then usually getting rekt myself due to trying to breach a fortified position alone]. But rather than seeing ai whitelisted [as that would screw over all the benefits an ai brings to station] I'd like to see more TEMPORARY BLACKLISTS - to have them act a little more like wiki says. 

Ex: "Room Y breach detected. Security response recommended" instead of "Four hostiles armed with X in room Y, red voidsuits, one rig". (See wiki, it genuinely recommends the ai acts like this)

I'll edit this post to be more coherent over the day - writing this in spare 15 mins before going to work.

[Carver]

1 hour ago, Naelynn said:

Ex: "Room Y breach detected. Security response recommended" instead of "Four hostiles armed with X in room Y, red voidsuits, one rig". (See wiki, it genuinely recommends the ai acts like this)

Let me tell you exactly why this doesn't work: The correct response to this is "Okay, what's in there?" "Okay, what are they equipped with?" "Okay, what do they look like?"

Letting the AI shorten the need to explain the details causes no harm, and only an idiot will go into a situation without such crucial information. Information gathering is the single most valuable part of having an AI, and the strongest reason to presently subvert one. If one wishes to avoid having information revealed, then plan ahead and be wary not to show all your cards (or just subvert the AI first).

[Jupiter Storm]

Thank you for the replies to this thread, I think it's worth mentioning that I, strictly speaking, don't want to see the A.I. whitelisted either? Merely, I am throwing the feelers out there by way of a potential compromise to avoid the 'other' thing that the rumours suggested.

Honestly, that stuff gets really scary - I know there are a lot of people who are really indifferent and only see A.I. as a tool and w/e, but for players who are passionate about it, and enjoy playing it, and want to make actual characters out of it, rumours can be extremely unpleasant to hear. When a head of staff straight-faced tells you, "Soon. I'm removing it. Don't worry, it'll still be available from Robotics." - what are you supposed to think? I know BygoneHero has had a similar experience to this and actually ended up arguing about it without being told "Okay stop, it was a joke dude."

I know that there exists a following of players who truly, truly despise the role, and who disagree on a personal level that it should even be a part of roleplay, but... I just can't wrap my head around that, I've always been down with a 'live and let live' policy. And as I said, Whitelist was a compromise for that, to add more wheat and less chaff. Not that I personally am really 'eager' to see it happen.

From what I recall, other balancing options were suggested as well, and maybe this is the place to discuss them while the topic is being looked at? I had some of my own:

- I know that Paradox wanted to draw down the number of cameras further to give a few more 'hiding places in plain sight', so to speak.

- Add a 3-minute or 5-minute delay for Alarms to appear in A.I. and Borg logs, but still make them immediately appear on consoles if somebody is bothering to manually check these, i.e. Engineering or Security.

- Only allow crewmembers to be tracked on cameras if enable their Tracking Beacons on their Suit Sensors, meaning that anybody who wants to use the A.I. as a "follow me and give me access" service, will get exactly what they want, and anybody who doesn't want to use the A.I. and prefers a bit of anonymity cannot be tracked in the conventional manner. Naturally Space Carp, Cavern Dwellers and Spiders can still appear on cameras normally, so that the A.I. can point them out to people and make Sec's lives easier clearing out mobs before they become a problem.

One more thing worth mentioning - if the idea of whitelisting makes people uncomfortable, maybe that's an issue with how whitelists are viewed? There seems to be an opinion in the community that 'whitelisting' a role inherently means you just won't get it? It doesn't, it just means that it's not immediately accessible to you unless you're passionate enough about a nuance of roleplay that you're willing to put in a little bit of effort. We certainly have enough active Lore Devs that you won't end up waiting weeks upon weeks for it. But like I said, I don't specifically 'want' a Whitelist. I'm merely throwing out potential compromises without completely changing a role that some of us truly enjoy.

8 hours ago, Naelynn said:

Furthermore: Ai is an excellent role to LEARN COMMAND. It is the only non-command role to have access to their channel, and can observe how things are to be handled. This is suuuuper important for people who are looking to whitelist for command. You can say that Warden is the 'Learn Command' role, but... not really, even though they have access to fax machines.

I have always said this, too. A lot of people who have Command Whitelists (won't name names here) have absolutely no talent for communicating and coordinating with others, no listening skills and seem to become easily daunted by rapidly developing situations.

Anybody who will ever claim that A.I. doesn't prepare you for keeping a cool head while being spammed with radio logs and multiple situations that require timely responses and coordination has either not played it, or hasn't been passionate enough when playing it to do the job properly.

[Shadow]

I wholeheartedly agree with carver, everything that I wanted to say he has said.

[Xelnagahunter]

11 minutes ago, Shadow said:

I wholeheartedly agree with carver, everything that I wanted to say he has said.

This.

While I would prefer to see a whitelist over removal, I don't think either should be an issue. I think enforcing good RP and objective planning are things we should be more worried about. If people think the antags having to make plans around the AI existing is hard, then perhaps they should turn off antag options until they learn how the AI operates, or even try playing a few AI rounds themselves. It's really simple to avoid cameras and to avoid the AI finding you when you are in front of one (unless they happen to scroll over you on the map and know how you're dressed). It's also simple to subvert the AI in a manner that most people would never notice for those doing stealth ops.

In all other purposes the AI is a huge help and even a teaching tool if they know how various departments work.

As said, if the options are whitelist or removal, I'd much prefer a whitelist. However I don't think either is a valid option really.

Edited August 29, 2019 by Xelnagahunter

[Dark1Star]

Time  and time again have I NEEDED the AI to do something for me, usually if there's a situation when there's no command and there's an AI, a lot of the time the crew rely on the AI. AI, open the armory, and could you send a cyborg to open the lockers, or AI, quickly open this door! They're in here getting dressed! etc etc. Having no AI, I can see it having a huge effect on the overall flow of a round.

[Arrow768]

To clear something up:
There are currently no plans by the development team to remove the AI or make them a thing that does not exist by default and explicitly needs to be built.

That said, I agree that there is a quality issue with some AIs, especially with those that try to simulate extensive "personality".
However, I do not agree that the way to go about this is yet another whitelist, as whitelists do not do much in terms of quality control. (See various heads of staff out there)
First and foremost they are a means of limiting access.

Since something needs to be done we are currently discussing if the issue can be resolved by being more aggressive with our blacklist (AI jobbans).
Once there is something more concrete, it will be published on the forums/discord.

Therefore, as there are no plans to remove the AI out of standard rotation (which seems to have prompted this topic) and because I believe that yet another whitelist will not be able to solve the AI quality issues I am voting for dismissal of this suggestion.

[JKJudgeX]

I have made or contributed to suggestion threads that discuss nerfing the AI's ability to track and spy on antagonists to no avail.  I've suggested highlighting cameras when the AI is looking through them, re-wording the wiki and rules relating to AIs to have them use their abilities with much more discretion in mind, and so on.  

That being said, I think this post would be 100% good for the game.  Most of us who have played for a long time know at least 3 or 4 AI players who are really good at playing AI with the restraint necessary to allow the round to continue to develop while being attentive enough to keep the AI's function useful, so those people should have no problem getting a whitelist for being AI...

I've played hundreds of rounds of antag here, and a very good percentage of them have been spoiled by an overzealous AI watching my every action.  The AI was never intended to be played as an omnipresent member of security that secretly reports suspicious actions (often only suspicious in a meta-sense, too, btw).  

I've referenced it before, that on the wikis of other servers, and commonly 3+ years ago that the playstyle of the AI is recommended to be as such:

Here's an example from the SS13 fandom wiki: 

Quote

 

"Absolutely Goddamn Terrible :
HAL9000 : DAVE JUST EMAGGED A DOOR HE IS BOLTED IN CATCH HIM QUICK.
Very Bad :
HAL9000 : Dave is the traitor.
Fair :
HAL9000 : Dave is breaking into the Captain's Quarters.

Better :
HAL9000 : Someone is currently breaking into the Captain's Quarters.
Mike : Who ?
HAL900 : Dave.

Best :
Mike : Where the hell did my spare go ?
HAL9000 : It is gone.
Mike : AI, do you know anything about this ?
HAL9000 : Yes.
Mike : Alright, who broke into my office ?
HAL9000 : Dave did."

 

On an HRP server, I would think that in reference to the above quote that we would expect the "Best" style of play - that is, before reporting anything that a crewmember has done that is not directly and immediately threatening someone's life, that the AI would need to be queried about it rather than just suddenly announcing to security or command what had been done.

This is nowhere near how AI is played, and I've had AI players go as far as to listen in to painstakingly set up private conversations via intercomms and report the contents of said conversations directly to security - just as one of many, many examples I, or likely other players, could give.  

Since that's the case, I'm in support of whitelisting, and, failing that option, complete removal of the role.

Also, I'm pretty sure MALF is near-universally disliked when it's rolled - by most non-AI players especially - because it's A) Pretty much single-antag, and B) 9/10 times played as stealth until Delta is reached.

That being said, I will also note that a long series of changes that empower antagonists have been made in the past several months/year - reduction of access to ballistic weapons to crew, removal of ERT calls during rev, nerfing of captain and HoS's baton, and some other things here and there - and I'll say that I think some of those decisions should be revisited at the same or near the same time to make sure that antagonist play is not too easy and still has to rely on a modicum of stealth and cleverness for dramatic antag-"win" outcomes.

I'm disappointed to see such a quick dismissal on this post.  I think it is very reasonably stated and written by someone who understands this issue innately.  Regardless of whether there were existing plans to do anything or not, there should be, as this little issue has been festering for quite some time.
 

Edited August 30, 2019 by JKJudgeX

[Carver]

Funnily, of your example, the truly best AI would be the 'Fair' AI, for being concise.

'Terrible and Very Bad' would only be fine on an MRP/LRP server, 'Better' is simply being vague. 'Best' is either subverted by the antagonist or utterly fucking useless.

Now, if you're having issues with AIs, as earlier I suggest playing around them. Having an obstacle will not 'spoil' your round, so for the sake of minimizing my response I will quote myself:

23 hours ago, Carver said:

They're one of the few credible threats left to antagonists to keep them in check, and I stand by my belief that the game should not be centered around the gimmicks and whimsies of antagonist players (unless they themselves were whitelisted).

If I may recommend a method for avoiding your AI issues? Spend telecrystals on an Agent ID and don't call attention to yourself. There are a plethora of answers to any such problem.

[Cnaym]

38 minutes ago, JKJudgeX said:

That being said, I think this post would be 100% good for the game.  Most of us who have played for a long time know at least 3 or 4 AI players who are really good at playing AI with the restraint necessary to allow the round to continue to develop while being attentive enough to keep the AI's function useful, so those people should have no problem getting a whitelist for being AI...

I've played hundreds of rounds of antag here, and a very good percentage of them have been spoiled by an overzealous AI watching my every action.  The AI was never intended to be played as an omnipresent member of security that secretly reports suspicious actions (often only suspicious in a meta-sense, too, btw).  

Pushing a bit back on that one. For some reason the good AIs leave after a while. One of the reason being security shouting at them to much.

Now JupiterStorm just caught the blunt of it. While most of us love her for being helpful, having deep knowledge, sending her borgs in time and building upon their strengths and so on... security felt that this AI just did not do the thing. There was rude behaviour ICly and OOCly. I'm honestly not sure how to fix a mentality problem since a slow but really well played AI would be a thousand times better than the silent door bolting one or the even worst all caps "VALIDS FOUND" Ai.

That said... yes, a whitelist would help a lot in terms of getting shitters out of the malf slot. A stricter ruling on getting rid of bad AI players also helps. A combination of sorts may be hard to get running, but I think temporary job bans for fuck ups should be in place  since the AI has so much responsibility to should... that would  also require stricter and easier to understand rules on the wiki. Maybe even a button that tells you to read the rules, you  have to agree to getting job banned rather quickly and only then can you enabled / chose that. Not sure how hard that would be to code, but it would create a soft barrier of "Hold on, am I ready for that?"

[JKJudgeX]

The problem with relying on "blacklisting" is that the issues that people bring up regarding AI very frequently are not against any rule.

There's no rule at all that says an AI can't lock you down and report it in full detail the instant it sees you break into a place, overhears you saying something incriminating, or watches you perform a cult ritual somewhere...  how would such a rule even be written?  And because of that, I think the whitelisting would be a good idea because it allows the playstyle to be more tightly controlled, and pressure put forth on AI to play and perform well - as a whitelist can be threatened, which makes every potentially questionable interaction a gamble to have your actions tested against the "good play" established by synth whitelist management (or whoever is in charge of dictating acceptable RP standards for AI).  

Though I tend to be cynical, I think that stating that a whitelist doesn't help anything as with current command or xenos in terms of quality is a bit off-target and not giving fair consideration to many of the players with those whitelists who work hard to portray their characters and participate in rounds in a positive way.  As a long-standing member of command, I have only a very small number of incidents of clashing with other command for not doing their jobs properly, while I have a host of examples of regular crew performing like absolute garbage and not even being sorry about it, so, take that as my testimony to the value of a whitelist, as I think they are as good, reasonable, and helpful as the quality of the administrators and moderators behind their issuance, governance and revocation.

[Jupiter Storm]

14 hours ago, JKJudgeX said:

There's no rule at all that says an AI can't lock you down and report it in full detail the instant it sees you break into a place, overhears you saying something incriminating, or watches you perform a cult ritual somewhere...  how would such a rule even be written?

The unwritten rule of narrative writing and progression of a story, which isn't the reason a lot of people play it would seem. I don't think it's even possible to write down a precise guideline to follow, since the art of making sure a story flows and a balance of power between good / evil is maintained is a nuanced art that takes time and careful nurturing to develop (which as the A.I., you can actually decide rather readily in a round, you're certainly an RP-enabler with a huge truck-ton of weight to throw in either direction).

I too am dissappointed to see a quick dismissal but it is what it is, since the rumours are as previously mentioned non existent even if a problem DOES exist which leads me onto...

 

22 hours ago, Arrow768 said:

There are currently no plans by the development team to remove the AI or make them a thing that does not exist by default and explicitly needs to be built.

Like I'd said earlier, somebody needs to acknowledge that there is a bit of a serious issue if people can hear these things and take them seriously. I and several other people have had members of staff and developers fully capable of effecting these changes straight-faced tell us that it's going to happen, and have gone on to argue the case and present some very valid reasoning behind their logic.

I understand that for those people, it's not a big deal, because the role is just a bit of a 'joke' to them, they don't really care because it's not the focus of their favourite lore-set, but for people who are actually passionate about the roleplay and the concept, rumours about big changes are a scary thing to hear and it is honestly not okay for any member of staff to go around saying that kind of stuff and lending it serious credence, and then expect it just to be brushed off?

I don't think I'm being unreasonable or rude in saying this, I'm hoping someone will see where I'm coming from here, which is the base reason why I even began performing my 'tests' and also posted this thread. >_>

 

14 hours ago, Cnaym said:

security felt that this AI just did not do the thing.

I actually specifically chose not to do the thing, as you know - as part of my experiments. And the result... Yeeeeeah... It goes to show one of two things, either...

#1 - Some members of the Security community - perhaps even large portion of it - are either too lazy to coordinate and scan for themselves, or lack sufficient understanding of the tools and personnel at their disposal in order to do so...

OR.........................

#2 - There is a hard-line acknowledgement by the Security community that without the A.I.'s assistance, they find it very difficult to perform their roles effectively. Which, once again, points toward just one of the A.I.'s massive suite of functions it offers that actively empowers the crew to deal with multiple, complex situations at once - something that members of Command would find exceedingly difficult when they CANNOT be in multiple places at once, no matter how effective they are as leaders.

And that's not even taking into account the other conveniences and roleplay experiences that the A.I. can provide the crew such as those mentioned by @Naelynn and @Xelnagahunter - particularly teaching newer people, acting as an encyclopedia of 'how do I do this thing I never heard of before'. Talking somebody with OOC knowledge through a process they do not know ICly, and performing most of the complex functions for them, in order to fill a dire need that the station lacks at the time (one of the few exceptions to this is surgery and the more practical side of engineering). (I highlighted this because it's a seriously overlooked function that a lot of Sci-Fi stories with A.I. characters include them doing. And it's so important, particularly when everybody else is dead and you want to stand some kind of chance at surviving without breaking the server rules.)

Or simply just providing other roleplaying platforms and a fixed point of neutral discussion for anybody who feels lonely and can't seem to click with the crew today.

Edited August 30, 2019 by Jupiter Storm

[ben10083]

9 hours ago, JKJudgeX said:

There's no rule at all that says an AI can't lock you down and report it in full detail the instant it sees you break into a place, overhears you saying something incriminating, or watches you perform a cult ritual somewhere.

There is a rule, it is called powergaming. Borgs get this alot when they would rush and try to baton da antag (with a hostage)

[LanceTheInvader]

Don't whitelist AIs. I often got stuck alone as an engineer apprentice of with very little help. Would have been unplayable if I didn't had an AI clearing the path before me. Saved me a great deal of time instead of hacking everysingle door I saw. Not having them whitelist gives the better chance of always having an AI available wich is often essential when understaffed.

[Rushodan]

19 hours ago, Arrow768 said:

To clear something up:
There are currently no plans by the development team to remove the AI or make them a thing that does not exist by default and explicitly needs to be built.

That said, I agree that there is a quality issue with some AIs, especially with those that try to simulate extensive "personality".
However, I do not agree that the way to go about this is yet another whitelist, as whitelists do not do much in terms of quality control. (See various heads of staff out there)
First and foremost they are a means of limiting access.

Since something needs to be done we are currently discussing if the issue can be resolved by being more aggressive with our blacklist (AI jobbans).
Once there is something more concrete, it will be published on the forums/discord.

Therefore, as there are no plans to remove the AI out of standard rotation (which seems to have prompted this topic) and because I believe that yet another whitelist will not be able to solve the AI quality issues I am voting for dismissal of this suggestion.

I don't really want to whitelist the AI either, would totally agree with being slightly more aggressive with the AI jobbans for AIs that only exist to scan the exterior and valid hunt. Contrary to popular belief security aren't allowed to do this either, so which should station synthetics?

That being said, I feel as if a lot of valid points for the AI being...well almost needed have been raised here which I will not repeat for the purpose of keeping this post concise.

Furthermore, as someone who plays antag a lot I actually like the challenge of having to stay off cameras if I want to do anything obviously illegal/against station regulations. Like, if I hear the AI making announcements I don't go "Oh look another valid hunting machine that will stop my rp", instead I try to plan around it - take a toolbelt in case I need to open a door that has been locked down etc. 

[JKJudgeX]

"Enforce corporate regulations" should not be part of its function unless specifically used to do so by those who actually hold those jobs, IMO.  Should it even understand trespassing?

[Jupiter Storm]

It's heavily implied in the A.I.'s lore and lawsets that it understands the concept of authorized and unauthorized personnel, and what unauthorized personnel are capable of doing with things they should not have access to.

Beyond this though, I am still a huge fan of mechanical nerfs that stop them from detecting antagonists so easily, unless they're actively tracking one who's really pushing their luck and doing a lot of naughty things at once in a short space of time.

Something that myself and Errant do quite a lot already is wait a certain amount of time before announcing alarms set off by antagonists (within reason, if it's early in a round, but if their gimmick is underway and it's a pretty loud murderboney one they don't get a free pass).