New AIs and Old AI Interactions

[Nist]

The roundstart Ai vs the creation of new ones

It is generally considered polite to ask permission from the existing Ai before going and making new ones. In fact, there are admins in the past that have actively acted in favour of such formalities, and consider them "the way to go" in this aspect.

Only recently did it come up in discussion to find that no such rule is actually codified.

The thing is, the Ai does not often have a workload that needs two or more people being an AI overseer. Just like with Janitors/Janitor cyborgs, it can feel to some Ai players like their toes are being stepped on.

It especially stands out to me as odd, that the creation of unsynced cyborgs for no reason is against the rules, but not Ai, which feels like an even greater deal. Specially considering that those secondary AIs will often take over the main AI's gameplay loop almost entirely, and it becomes very quickly a very frustrating competition between two players to see who clicks the button to open door quicker.

It also bolsters an unhealthy habit from some Sci members that indirectly use AIs to grief the main AI, by creating multiple other Ais over multiple rounds (Something I've been a victim of in the past to the point that I quit the game for a while over building frustration and lack of enjoyment). It is in my experience a 95% chance that such Ais will go sideways very quickly and become not only a nuissance to the round but often step over the main AI's tasks and effectively act as a passive round removal.

I think because the existing Ai stands the most to lose by the random creation of new Ais, it is fair for them to have actual authority over this one specific area and let it be the one who calls the shots on the creation of new AI cores for ON STATION CREW (With the exception of confirmed Malf/dead AI). This would not apply to Lavaland, Whiteship, Charliestation & other AI types that are designed around ghost roles.

[technokek]

You should not create a second AI without a clear purpose for it. This should fall under the same rules as unsyncing borgs for no reason.

[Striders13]

Yeah, having a secondary ai always sucks. But since modsuit AIs exist it doesn't happen nearly as often.

[zxaber]

I play AI (and silicon in general) quite a lot, so I have an opinion on this.

New AIs randomly showing up are rarely an issue. It is, I suppose, polite to let the existing AI know, but I'm against any rule that prohibits new AIs without "permission" from the station-start one. Malf/Subverted AIs can have it somewhat harder (as there's no easy way to send messages to synced cyborgs when binary is compromised), but it's not the end of the world. Rival AIs can be killed, and synced borgs can be advised through backroom holopads. More tools could help ease the pain, but we do have methods in place if you're creative. (I will admit that a notice for "New Connection" to binary whenever a new AI/borg is made would be nice, but I digress)

As a general recommendation, I usually tell new players that want to learn AI to go become a mid-round AI through robotics. This allows them to lean on the station-start AI for help and guidance, and lets them play with the controls without having to worry about being flooded by law-2 requests.

Law 2, grant me permission to make a new AI

[Djaubb]

It depend of the rounds but yeah, lately I often see the roboticists making a second AI (I swear it's just because they prefer to do ANYTHING instead of building cyborgs) without asking me, the shift start AI, anything. They don't ask RD/Captain either, but hey it's Terry.

What's annyoing it's that the seconday has EXACTLY the same capacities (aside it's not "protected" in a SAT); I became secondary AI thanks to a tot, and I was so sad for the main AI because I was doing a lot of shit (even in their sat) and were not aware of it until it was too late and even then they didn't know it was my doings, I think they didn't even know there was a second AI.

So yeah, shift start AI should have some access/interactions over the newer AIs; like, shiftstart AI is issued by Nanotrasen, has an AI sat securised and all, but no. It's equal to AI built by two nerds in their man-cave with monkey's meat as floor tiles. Why???

And I won't speak as of when there is 3 or more AIs...

And as OP said in their post: it happens that the secondary AI take over ALL the work. I'm the shiftstart AI but I have nothing to do because the second AI seems to be an AI main that for an unknown reason didn't start AI but still wanted to be (I always put AI on the lowest priority).
So yeah, that's when it gets annoying. Imagine if HoP could promote a new captain without asking permission and without repercussions...

What's more it's that most of the time the situation doesn't require to have a second AI, they build one just because they can.

[Ryusenshu]

I agree with everything zxaber said here
There is no need to add a policy for this

New Connection message for binary might be neat though, of course if its without location info

[Nist]

Rival AIs can be killed, and synced borgs can be advised through backroom holopads. More tools could help ease the pain, but we do have methods in place if you're creative. (I will admit that a notice for "New Connection" to binary whenever a new AI/borg is made would be nice, but I digress)

I've heard plenty of "It would intrude upon Rule 1 to remove people for X reason" arguments from people when I argue that Rival Ais can be killed, so it ultimetely falls as ever to the admin's personal interpretation on the cases that I've heard and been a part of. Codifying the correct response would go a long way to help newer AI players that are round-start deal with these situations.

As a general recommendation, I usually tell new players that want to learn AI to go become a mid-round AI through robotics. This allows them to lean on the station-start AI for help and guidance, and lets them play with the controls without having to worry about being flooded by law-2 requests.

True! Midround AI can be great to learn the basic functions, having taught some myself during my time on Terry and later Manuel, it's always good to see midround AI being used for learning purposes. However, Ais that simply drop a message like "I'm new" or "I'd like to learn X things" are very rare, compared to silent players that take over all your workload.

[terranaut]

imagine you're the HoS and every greyshirt could decide to turn another greyshirt into the HoS, and instead of that guy turning into fine red mist the moment he walks past the warden into the armory, he has the same permissions and rights as you entirely legally and if you don't like it, well sucks to be you i guess. extra ais are mega cringe
(i remember when nist would force extra ais when i'm playing, funny how the turn tables)

[Nist]

(i remember when nist would force extra ais when i'm playing, funny how the turn tables)

funny how the tables turned indeed, but good to see you around still terra.

Ultimately, the issue is that I can't really do anything without it being codified until waiting for shit to go horribly, horribly wrong and then act and delete out of existance. It's something that has hella implications for my play sessions, and I damn well know I'm not the only one that feels this way.

[Cobby]

you can unsync borgs you just have to have a valid reason per rule 4 as the more practical use is to preemptively btfo an AI thats malf or subverted.

I could somewhat understand that logic also being somewhat applied to an AI but it seems kinda lame if you could only build an additional AI if the other broke or something (you wouldnt really want to add another to the mix just for the sake of different lawsets between the two idt as one would want to kill the other).

Codewise binary should be separated between master AIs and/or AIs should be treated like borgs in the fact they start out sync'd to the mother AI so its all a cooperative (or at the very least the "veteran" AI is always trumping the newer one in power dynamics).

Until then though I personally dont think its an administrative problem unless they outright admit theyre doing it for antag hunting, its more gated than say a HOP who could job-flood a department.

[Timberpoes]

It also is an instant valid tell for an antag if they create AIs without the appropriate permissions when doing so would ordinarily be against the rules.

[mstachife]

I feel any sort of actual ruling here would unnecessarily limit interesting situations that can occur. I like the way it works out normally now with command and security probably kicking in the door if you make one without authorization, but it not being expressly against the rules.

[terranaut]

It also is an instant valid tell for an antag if they create AIs without the appropriate permissions when doing so would ordinarily be against the rules.

good
why would a random crewman build an extra ai with the power of stealthily killing the station ai and then subverting the station
space all extra ai builders

[Nist]

I feel any sort of actual ruling here would unnecessarily limit interesting situations that can occur. I like the way it works out normally now with command and security probably kicking in the door if you make one without authorization, but it not being expressly against the rules.

I've not had interesting situations with other AIs on station, it's consistently been a whole removal of interest because the other AI practically deletes my job if they have good ping, while they silently sit and don't interact with anyone in borgchat to even attempt organizing.

It also is an instant valid tell for an antag if they create AIs without the appropriate permissions when doing so would ordinarily be against the rules.

Then the antag can create it in maintenance like they already do with literally everything else AI-related (See: Uploads), which are also treated as instant valid calls if done without good reason.

[Not-Dorsidarf]

Having a second AI built fnr as a regular AI is either moderately annoying or a fun bit of companionship having someone to chat to who isnt always busy like borgs.

Having a second AI built fnr as a rogue / malf AI is an immediate round ruiner.

[dragomagol]

As long as you're not doing it to meta the original AI ("check if the main AI is malf"), which is already against the rules, we don't see the need to codify this.

Headmin Votes:
Dragomagol: Agree - I've had some fun experiences as malf AI against a regular station AI, and I've never felt that a second AI was overtly terrible to have unless it was doing something that both of us got blamed for.
NamelessFairy: Agree
RaveRadbury: Agree