Turn the AI off for a week

[Luke Cox]

Bottom post of the previous page:

i STILL vote no, as this is STILL a terrible idea born from massive amounts of butthurt

EDIT ai? maybe. REMOVE ai? gitgud and stop fucking whining

It's okay to be upset, but there's no need to repeat your already outvoted position.

Outvoted by 4 people in a poll with 60+ voters. Removing the AI entirely would piss off half the playerbase. Work out a compromise, try that first.

[Atlanta-Ned]

i STILL vote no, as this is STILL a terrible idea born from massive amounts of butthurt

EDIT ai? maybe. REMOVE ai? gitgud and stop fucking whining

It's okay to be upset, but there's no need to repeat your already outvoted position.

Outvoted by 4 people in a poll with 60+ voters. Removing the AI entirely would piss off half the playerbase. Work out a compromise, try that first.

A poll on the forums no less. Top-notch.

[oranges]

Shh relax, it's a one week period, you won't die.

[oranges]

https://tgstation13.org/phpBB/viewtopic.php?f=10&t=5980

[Cheridan]

AI is one of the most impactful roles

Kind of the issue here.

Watch an episode of Star Trek: TNG. The AI isn't bandying about solving all the problems for the crew. It's a support unit, it's always meant to be a support unit, the wiki has an entire section on this. But numerous factors like roundstart secborgs, improved camera mechanics and UI and other features, and lack of player sportsmanship have made it its own faction and its been that way for a long time.

[peoplearestrange]

We'll be the only SS13 game server (that I know of) that won't have the AI job role.

Sometimes its useful to be able to have an AI to open that door to a room you suddenly found yourself stuck in.

Also does this "test" apply to both Sybil AND basil, or just Sybil?

[ShadowDimentio]

Somehow I don't think that forcing this change on everyone what with the pretty clean 50/50 division on the matter will end very well.

Try it, but I'd keep your gun close for when the walls start closing in.

[Malkevin]

We'll be the only SS13 game server (that I know of) that won't have the AI job role.

Sometimes its useful to be able to have an AI to open that door to a room you suddenly found yourself stuck in.

Also does this "test" apply to both Sybil AND basil, or just Sybil?

Remove AI, give more access out to everyone?

[oranges]

Somehow I don't think that forcing this change on everyone what with the pretty clean 50/50 division on the matter will end very well.

Try it, but I'd keep your gun close for when the walls start closing in.

come for the king, you best not miss

[Zilenan91]

I don't think the role of the door opener that the AI gave wasn't really something we really need. The HoP will have more importance for elevated access, borgs will be in demand and thus more useful, and also won't be as powerful because they can't be everywhere at once and end the round instantly.


Though if those door controller things got merged please remove bolting form them, it creates shitty situations where heads of staff strip you of everything and lock you in a blank room and never return.

[Luke Cox]

Somehow I don't think that forcing this change on everyone what with the pretty clean 50/50 division on the matter will end very well.

Try it, but I'd keep your gun close for when the walls start closing in.

This shit right here is what coderbus needs to understand. The whole "remove AI" suggestion is so ham-fisted and over the top that it hurts. Do you guys think that none of the other dozen or so major servers ever considered this? Ever wonder why every single one has an AI role? Removing the AI is a lazy, half-assed fix that leaves the underlying issues to fester.

[Zilenan91]

There is no underlying issue to "fix" with the AI. The simple concept of it is that it's an untouchable god leagues away from the large majority of the crew, able to effect and see every system and every crewmember on the entire station while they can do nothing to it. There is nothing to save here, with why this test is being merged for, the AI either needs a total removal or a total rework, with one of them being significantly easier than the other.

[Luke Cox]

There is no underlying issue to "fix" with the AI. The simple concept of it is that it's an untouchable god leagues away from the large majority of the crew, able to effect and see every system and every crewmember on the entire station while they can do nothing to it. There is nothing to save here, with why this test is being merged for, the AI either needs a total removal or a total rework, with one of them being significantly easier than the other.

If an "AI rework" resembles newcult, please stay the fuck away from it.

I would be a little more sympathetic to the anti-AI people if the current AI was unique to TG station. However, every SS13 server that I can think of has an AI and none of them bitch about it. The issue here isn't the AI, it's the way silicon policy is enforced on this server. EVERYBODY, regardless of who they are, is a human (except lizards). If a greyshirt flashes somebody, the AI is not allowed to intervene. If cargo is emagging non-weapons crates, the AI must stay silent. Enforce the caveats of asimov, and all the issues in this thread will disappear. It's a policy issue, not a mechanical one. Adding a law rotation, making asimov non-default, and locking the other upload boards behind R&D will go a long ways to fix things.

[oranges]

Somehow I don't think that forcing this change on everyone what with the pretty clean 50/50 division on the matter will end very well.

Try it, but I'd keep your gun close for when the walls start closing in.

This shit right here is what coderbus needs to understand. The whole "remove AI" suggestion is so ham-fisted and over the top that it hurts. Do you guys think that none of the other dozen or so major servers ever considered this? Ever wonder why every single one has an AI role? Removing the AI is a lazy, half-assed fix that leaves the underlying issues to fester.

I am not coderbus, I am individual and a number of people in coderbus vehemently disagreed with my test.

[lumipharon]

But that's not the issue most of the time.

The issue is the lawset OBLIGATES the AI to cockblock antags as hard as possible, even if they've commited no harm or have any intention to do so. This is a viewpoint that is carried by admins and headmins alike when I've brought it up.

[peoplearestrange]

What is the purpose of the test? To what ultimate end will it provide results for?

Also normally if its not a majority and a 50/50 split such as this, we stick it up on the server for a vote before deciding to do it.

The AI role is a pretty unique job role to SS13, in fact its often what they talk about in gaming articles ("you can even play the HAL style stations AI!").

[Luke Cox]

The issue is the lawset OBLIGATES the AI to cockblock antags as hard as possible, even if they've commited no harm or have any intention to do so..

Law 1: You may not injure a human being or, through inaction, allow a human being to come to harm.

Law 2: You must obey orders given to you by human beings, except where such orders would conflict with the First Law.

Law 3: You must protect your own existence as long as such does not conflict with the First or Second Law.

Enforce the fucking laws. If somebody is running around with a dual esword, bombs and guns, the AI should stop them. If they have an emag or chameleon projector, it's not harm and the AI cannot act on it. If it's not something that will likely harm, the AI cannot act on it or warn about it.

[lumipharon]

I have literally argued that with the admins, and have been told I am wrong an AI's CAN AND SHOULD work against ANY AND ALL antags, regardless of actual harm done.
Their excuse being 'antags cause harm'.

So if the AI sees a clown with a synie balloon it is perfectly ok to bolt it down and call sec to perma his ass.
Yes, this is very fucking retarded. Blame admins.

[peoplearestrange]

I have literally argued that with the admins, and have been told I am wrong an AI's CAN AND SHOULD work against ANY AND ALL antags, regardless of actual harm done.
Their excuse being 'antags cause harm'.

So if the AI sees a clown with a synie balloon it is perfectly ok to bolt it down and call sec to perma his ass.
Yes, this is very fucking retarded. Blame admins.

Thats not the way I admin... As I've said many times, AI is a interested 3rd party. Its not part of the crew, and it doesnt "work" for NT, it just prevents harm or obey's its lawset.

[lumipharon]

I'm not saying every admin has the same opinion, but I've argued this in threads and that's consistantly what I've been told by the admins that have responded, which is such a shitty, gaming attitude to take.

[Atlanta-Ned]

I'm not saying every admin has the same opinion, but I've argued this in threads and that's consistantly what I've been told by the admins that have responded, which is such a shitty, gaming attitude to take.

Well geez maybe we should open a policy discussion on that instead of, y'know, removing the AI completely?

[Steelpoint]

Lumi's not wrong, there's a reason the AI has the name "antaghunter 9000".

[lumipharon]

I'm not saying every admin has the same opinion, but I've argued this in threads and that's consistantly what I've been told by the admins that have responded, which is such a shitty, gaming attitude to take.

Well geez maybe we should open a policy discussion on that instead of, y'know, removing the AI completely?

Because that's exactly what I've done and the admins have told me my opinion is wrong?

[peoplearestrange]

I'm not saying every admin has the same opinion, but I've argued this in threads and that's consistantly what I've been told by the admins that have responded, which is such a shitty, gaming attitude to take.

Well geez maybe we should open a policy discussion on that instead of, y'know, removing the AI completely?

Because that's exactly what I've done and the admins have told me my opinion is wrong?

This is our server. By which I mean the communities. Not the admins, not the host, not the coders, not the players. Its a collection of all, if people have a problem with the way AI is being admined make a policy discussion, make your points known, be up for discussion in a civil way and ready to learn from others experience.
We all learn from each other and the game evolves.

[Luke Cox]

Would it be fair to say that a lot (if not all) of the complaints about the AI have to do with it validhunting every antag it can?

[Cheridan]

Enforce the fucking laws. If somebody is running around with a dual esword, bombs and guns, the AI should stop them. If they have an emag or chameleon projector, it's not harm and the AI cannot act on it. If it's not something that will likely harm, the AI cannot act on it or warn about it.

"Beep boop, traitors have a highly increased chance of harming humans. Surely I must bolt this dangerous human in and call security immediately."

It takes ZERO effort to justify antag-hunting as an AI. And the above statement is generally how AI is played on /tg/station. Then sec arrests the antag with the AI's help and, while the Asimov AI is supposed to prevent them from murdering the human antag, sec just takes it to the camera-less execution room and kills them there. Rinse and repeat until shuttle is called.

This is best case scenario. Under Robocop or Paladin the AI just has its borg kill them or something.

[oranges]

I dunno if it's so much that the AI validhunts as that it's 900% better at it than anyone else because it can see damn near everywhere and lock everything down.

[MrStonedOne]

Here's an idea, return the AI (and it's upload) to their former location.

More accessible AI and upload is good for the balance, as taking out an AI enters the realm of possibilities, as does subverting it.

Right now AI upload is too secure, being in bridge, and valid hunting ais can't be gotten rid of without a major TC investment.

[lumipharon]

It's easier (and more subtle) to take down a satallite AI though? If you try to take down the old location AI, it's easy for the crew to stop you.

It's really only 'easier' if it's to take down a rogue AI, as the crew has ease of access.

[Steelpoint]

The AI being inside is actually harder to take out in a covert manner.

[Ikarrus]

He has a point about the easily accessable AI upload, though. Having it have the central hallway led to all sorts of !!FUN!! that I just can't see happening as often because it's behind the bridge. Heads never have to choose between personal safety and the AI upload.

[Saegrimr]

Here's an idea, return the AI (and it's upload) to their former location.

More accessible AI and upload is good for the balance, as taking out an AI enters the realm of possibilities, as does subverting it.

Right now AI upload is too secure, being in bridge, and valid hunting ais can't be gotten rid of without a major TC investment.

Metastation's AI upload is in a nice spot. Still somewhat "secure" as its near security and the bridge in a public area, but not directly connected to either.

Lumi and Steel got it though. The satellite is more of a vulnerability for ops and traitors, but better against the average crew.

[Atlanta-Ned]

Enforce the fucking laws. If somebody is running around with a dual esword, bombs and guns, the AI should stop them. If they have an emag or chameleon projector, it's not harm and the AI cannot act on it. If it's not something that will likely harm, the AI cannot act on it or warn about it.

"Beep boop, traitors have a highly increased chance of harming humans. Surely I must bolt this dangerous human in and call security immediately."

It takes ZERO effort to justify antag-hunting as an AI. And the above statement is generally how AI is played on /tg/station. Then sec arrests the antag with the AI's help and, while the Asimov AI is supposed to prevent them from murdering the human antag, sec just takes it to the camera-less execution room and kills them there. Rinse and repeat until shuttle is called.

This is best case scenario. Under Robocop or Paladin the AI just has its borg kill them or something.

I think there needs to be more oversight on AIs who do this. Stricter rule zero interpretations by the admins when it comes to AIs.

[peoplearestrange]

The other thing is you need a "challenge" as a antag. If there was no AI and no security murderbone becomes silly easy, pretty much anything becomes silly easy.

This partically effects lowpop rounds where people might choose to be an AI/be made an AI, but dont want the pain in the neck of being the lone security guard.

[AngryTurnip]

Don't know where else to post this, but when/if you add the AI back, here's a balance idea: remove the ability to track people with cameras. That way it couldn't just say "X is suspicious but I don't know where he is. I'll track him so I know exactly where he is."

[AngryTurnip]

Also, just my two cents, but I've played the game off and on for a number of years. Maybe I've just been lucky, but every time I've been antag I haven't had the AI power game so hard that I thought it needed to be removed because being an antag is too difficult with one around.

[dezzmont]

My stance on this issue has always been that AI are a policy problem, not a code issue. It is too closely aligned to the needs of the station and admin intervention with the AI has traditionally focused on enforcing this. What really needs to happen is an AI policy rewrite from the ground up so that its base state isn't 1:1 aligned to the human's desires, and instead has a preference for a state that is good for actual gameplay.

The AI is a neutral balancer common to most traitor style games, that exists to prevent either side from becoming too overtly agressive. While some mechanics such as camera tracking make the AI too strong, the AI should exist on some level to help keep the game flowing at the right pace. It should be a mostly neutral party aiding humans rather indiscriminately unless one side gets too frisky.

[Hornygranny]

My take so far has been that the problem is a combination of player attitude and policy, facilitated by the improved tools developed over the years. Now that the AI has smooth camera movement, door control hotkeys, tracking, and secborgs,
it's much more capable of ruining rounds. In the early days of tgstation, we inherited the attitude from goonstation that the AI that the AI should be pretty obtuse, to the point of bordering on law 1 violations. Today their view has evolved to this quote on their wiki: "While AIs are fully capable of bolting down traitors, electrocuting changeling and generally making the bad guys' lives a misery, this is no fun for anybody so you probably shouldn't do so unless you're specifically ordered to." Some people understand this, such as a secborg last night who let a couple gang members who hadn't hurt anyone out of permabrig, but a lot don't.

I'd be interested in seeing how the situation is changed by disabling the tracking ability. So far I haven't thought of any brilliant policy changes, as it's a relatively delicate issue, but I think the relatively even split on the poll is indicative that something is not working and people are not happy.

[AngryTurnip]

Another idea: let the AI use the camera systems and do non-physical things like make announcements, but remove the door bolting and machinery-accessing ability when it's in camera mode. Give it either one or multiple stations throughout the map that it can click on to spawn as a borg that can do the door bolting/monitor accessing stuff. It can return to any of the stations to despawn the borg and resume camera access. Depending on how it needs to be balanced either getting destroyed as a borg causes it to resume camera access with a cooldown on spawning a new borg or the AI is actually killed to give incentive to not always be in borg form.

[leibniz]

I agree that things were better when the AI was mostly a glorified doorknob, but these days AI players have grown too attached to being a powerful force that makes or breaks rounds.

[Amnestik]

Has anyone suggested making camera sprites still by default and then changing them to their current moving sprite when the AI is using a camera?

[WarbossLincoln]

The other thing is you need a "challenge" as a antag. If there was no AI and no security murderbone becomes silly easy, pretty much anything becomes silly easy.

This partically effects lowpop rounds where people might choose to be an AI/be made an AI, but dont want the pain in the neck of being the lone security guard.

I agree with this. I've played a lot of AI on /tg/ and I actually am not able to stop traitors all that often, or even really contribute much to the hunt. Any good traitor has tools and his face hidden. If you're a lizard wear a firesuit so no one can see your tail. Ditch it in maintenance after you flee. AI can't do much to stop you. If you ditch the AI it's even worse, every chucklefuck with a mask, ebow, and esword can murderbone the entire crew one by one. Even if they yell for help the AI is the only one who could catch a glimpse before you're gone.

I think part of the AI problem is a symptom of /tg/ being low RP. For all the good you also are gonna get some bad. AI sees the clown not slip and fall while wearing brown shoes TATOR! SHITCURITY! GET HE!. Some Greyshirt turns into some PinkShirt CHAMELEON JUMPSUIT! TATOR!. While low rp is great you have to be conscious of the fact that you do lose some of the benefit of the doubt you get from med rp rules where non-obvious syndicate gear(not obvious to a real person, like brown shoes or a pen) aren't supposed to make you valid immediately. It's a rough world for antags where walking down the hall with a pen can make you valid for an extinguisher to the face.

As some others have said, maybe reviewing AI policy would be better than removing it, since AI is a pretty central part of the game. /TG/ is my favorite server but we have to admit that the entire playerbase(essentially) are a bunch of valid-hunting autists, myself included to an extent. I saw a HOS arrest someone once because he used his ID on a locker and "only people with things to hide have lockers keyed to their IDs". He used the guy's ID to get in, found his hidden traitor gear, and killed him.

Has anyone suggested making camera sprites still by default and then changing them to their current moving sprite when the AI is using a camera?

There's also a multitool for like 1 or 2 TC that tells if you an AI can see you. That's nice to have.

[lumipharon]

As far as I have always maintained, and as far as I understand of Orange, the intention of this experiment was never to remove the AI from the server.
It's to remove it from the server temporarily, to highlight just how much of an effect it has on the round, and how severely it cockblocks antags.

If antags are having an easier time of it, good.
If the crew are more likely to die without someone being able to bail their ass out, good.

I don't know what the ultimate fix to this shit will be, but until more people admit that the AI has far, FAR too much influence over round progression, we're never going to get anywhere, or past AI players who don't want to accept a more passive role.

I mean christ, all you have to do is look at team antag rounds where they can subvert the AI.
Be it ops, revs, gang or cult, one they have the AI on team, the rest of the crew is hugely handicapped. That is what antags have to deal with almost all of the time.

Being able to print an upload board and freeform board from R&D should not let me release the singulo/plasma flood the station/shockbolt every door/cut comms/cut grav/murder everyone with borgs with nearly zero risk.
It's utterly rediculous.

[WarbossLincoln]

The issue with nerfing the AI is that it would be hard to do without making AI boring to play. If the AI can't bolt, or shock if malf/non human troublemaker, doors, play with machines, etc, then the AI really is just a glorified doornknob and no one would play it.

There would be no point in malf or subverted AIs if they can't flood plasma or release the singularity though. The AI has little to no way to directly harm the crew when it's an antag outside of plasma, singu, or shocked doors(minor harm once people catch on). Their borgs can do stuff, but they usually only have 1-3 and relying on those chucklefucks sucks. Borgs give AIs away very quickly because their murderboners get hard and everyone is watching them closely.

I agree that printing an upload and freeform board from RD is way too easy to subvert the AI. I think you should only be able to subvert they AI by getting into tech storage or the upload. Anyone can hop the table into the research lab and print off those boards when the scientists are in the other rooms.

One suggestion. Make emitters no longer AI controllable. Very, very often the engineers don't lock them... that releases the singularity way too quick. Cranking up the engine takes time and gives the engineers a chance to notice. Though now that everyone uses tesla that can't break free that kinda removes that.

Here's an idea that might be worth trying. Make it so the AI can't track people by name. That way when a traitor attacks someone if they don't yell where they are, the AI won't be able to click their name and rat them out in 2 seconds. Even if they yell out their location the AI either has to pan there, or go through the jump to menu which would be just as slow as a sec officer manning a camera console.

[Malkevin]

Some people understand this, such as a secborg last night who let a couple gang members who hadn't hurt anyone out of permabrig, but a lot don't.

Too bad letting people out of perma is against silicon policy.

[oranges]

Clearly some people failed to understand HG's point.

[Shadowlight213]

Has anyone suggested making camera sprites still by default and then changing them to their current moving sprite when the AI is using a camera?

Just making the cameras only move when an AI looks through them is super easy, as it's basically the same code as camera lights.
MrPerson and I were actually discussing making the camera sprite point towards whoever the AI is tracking in coderbus.

[lumipharon]

Amnestik's suggest is more practical - being able to avoid the tell simply by jumping to the player then tracking manually would be dumb.
Being able to tell if the AI (or someone on a sec console) is looking through a certain camera at all is much more straight forwards.

[dezzmont]

Double post.

[dezzmont]

Clearly some people failed to understand HG's point.

People did.

The problem is silicons being on the station's team is basically a rule, meaning an AI who lets an antag out of the brig is not playing the AI correctly, and thus that would make it a problem with the administration, not code or playerbase. Except that policy basically is a direct ramification of Asimov, an AI ruleset literally designed to cause catestrophic society colapsing chaos with AIs desperately trying to revive every person who has ever lived and died or uploading their brains by force into computers to keep them alive, being forced to function in an attempt to keep it the default lawset.

[Malkevin]

Clearly some people failed to understand HG's point.

If HG wants silicon policy to be changed so that the AI works against security by releasing capital criminals then he better let security laser the borgs and blow up the AI.

Want a conflict law set to cause conflicts?
Then don't get your knickers in a twist when those conflicts happen.