Utter lack of AI

[Zilenan91]

Bottom post of the previous page:

Basil isn't a fun place to be. There's a looooot of pubby griefers and the metafriending is still terrible.

[Shad0vvs]

Basil isn't a fun place to be. There's a looooot of pubby griefers and the metafriending is still terrible.

Oh wow, so I guess the AI isn't a big problem that should be removed if you're not willing to deal with that to not have it. :^)

[lumipharon]

Basil isn't a fun place to be. There's a looooot of pubby griefers and the metafriending is still terrible.

Oh wow, so I guess the AI isn't a big problem that should be removed if you're not willing to deal with that to not have it. :^)

Why not do it all basil instead so no ai can be with all the unrobust people?

[Shad0vvs]

Basil isn't a fun place to be. There's a looooot of pubby griefers and the metafriending is still terrible.

Oh wow, so I guess the AI isn't a big problem that should be removed if you're not willing to deal with that to not have it. :^)

Why not do it all basil instead so no ai can be with all the unrobust people?

My point is the AI isn't obviously the main problem we should be dealing with if its not preventing people from playing, while other things are.

Your point is ??? I like to play with different types of players or something.

[oranges]

Does anyone know where the other thread disappeared to?

[oranges]

Huh it moved to general chat, I thought for sure I started it in policy subforum

anyway thanks.

[Steelpoint]

I like the idea of doing what we did with Sec and basically cracking down VERY hard on Silicon players (more so AI's) to remove any shitty behaviour, as well as trying to move the Asimov lawset to a more neutral setting.

The AI is a unique aspect to the game, no one will deny that. But it needs to not be so heavily biased against the round's antagonists. High level antags aside such as Ops or Wizard the AI should not really get involved in mundane affairs unless things escalate hard.

The problem we had with the AI was that it simply was too easy for the AI to shut down many antagonists, and the AI's presence on in the round made it something that could not be avoided and had to be dealt with in some way.

Sadly there's a massive difference between dealing with a proactive Security Officer versus a proactive AI. The Officer can only be in one place at one time, the AI can be anywhere and everywhere at once. Not to mention the AI is far better suited to hunting down hostile forces than the officer can be.

[GeeElAge]

• You can see identification data (what's written on ID cards) but not characters themselves so that if someone equips the Captain's ID card, you see an image of the Captain generated from I assume the same place as the pictures in the crew records; you're seeing data in the way a computer does but abstracted for a game. The ID card says everything you need to know, unknowns could pop up as a black silhouette with a white question mark on their body for further emphasis and everything is cut and dry given current default laws.
• To see things in detail, interact or talk you need to "appear". If someone installed a computerised human in your workplace or school today, would you like them to do things without you seeing or would you like them to let you consciously know they are there and acting? In practice the former would be better for efficiency but the latter would likely be lobbied for out of concern over malicious action or to give ahead of time alert to prevent accidents, so why not force the AI to find a holopad and to use it before they can do most actions that aren't just observing information.

How would this work with faking identities and all? Can't really display an image of a crewmember, if they're not on the manifest.

[Cheimon]

I honestly don't understand where things have moved to now; I didn't get to play during the time of no AI and I haven't played since.

I'd like to see us move further away from the Asimov mindset such as going to something like:

• Do no harm.
• Stop currently occurring harm.
• Assist humans.

That's a decent lawset. It's simple enough to understand very quickly and it is quite different enough from asimov to stand on its own.

Code it, add it to the upload, and I'd happily use it if I was playing captain or HOS.

[Drynwyn]

Someone somewhere around here said that the problem is basically that AI's veer towards action over inaction, because inaction is boring for the player. This is a damn good point. Here's what I perceive to be happening: AI players want to act- they want to assist SOMEone, whether that is the station, the antagonists, or just Joe the Autism Forter.

They veer towards assisting security because

a) Security is easy to find and easy to help. They know who the sec officers are, and because they have Sec radio, they know what they need.
b) Antagonists are HARD to find and help, because they're used to needing to dodge the AI.

[Cik]

huh

you know what, that's a good point.

you know, what you could do is give traitors a 1tc key that gets them onto AI private, so they could ask for things, but without all the power of listening on binary.

it's a little risky, because unless somehow comms is changed it will reveal your true name, however if you're not planning on murderboning it may very well be worth the risk of being outed by validbot9000

[Drynwyn]

huh

you know what, that's a good point.

you know, what you could do is give traitors a 1tc key that gets them onto AI private, so they could ask for things, but without all the power of listening on binary.

it's a little risky, because unless somehow comms is changed it will reveal your true name, however if you're not planning on murderboning it may very well be worth the risk of being outed by validbot9000

I dunno if it would work without both a policy change and a large effort to make admins aware that AI's

a) Don't have to be aware IC of the fact that antagonists likely have an objective that requires killing.
b) Don't have to treat antagonists who haven't harmed anyone as harmful.

Currently helping antagonists as AI, while it generally won't get you banned, will tend to get the admins breathing down your neck with a "just give me a reason" attitude.

[Cik]

i do it all the time, and i don't think i've ever been bwoiked over it

but if you're right it should definitely be changed because that's dumb considering exclusively helping the crew is what everyone these days is upset about

[Atlanta-Ned]

I dunno if it would work without both a policy change and a large effort to make admins aware that AI's

a) Don't have to be aware IC of the fact that antagonists likely have an objective that requires killing.
b) Don't have to treat antagonists who haven't harmed anyone as harmful.

Currently helping antagonists as AI, while it generally won't get you banned, will tend to get the admins breathing down your neck with a "just give me a reason" attitude.

This is the policy we need to implement.

[Wyzack]

Maybe we should tral that after the no-ai trial and see if it helps, what do you think citrusman?

[dezzmont]

Giving the traitor a cheap or free way to talk to the AI without anyone knowing would be neat, especially because it would make the concept of a neutral AI really overt and obvious, the fact that a traitor can whisper the AI means that the AI should, on some level, care about this person on a level beyond beating the crap out of them.

Frankly I think the AI shouldn't care if a traitor killed someone that much, the focus on harm is almost out of place in SS13, though obviously the AI itself shouldn't go around killing people by default. I aways liked the idea of the AI being a centcom automated overseer who is more concerned with keeping the trains running on time than keeping people safe, because it creates scenarios where the AI will directly work against security, because it will now attempt to stop distruptive levels of valid hunting, and also adds a baked in reason for the crew to not constantly tamper with the AI for no good reason besides the fact they want to do something the AI doesn't want them to do.

As for the inactive AI bit, perhaps it suffers the same problem as engineering and there should be more passive, low benefit jobs that the AI can aid with.

[TheWiznard]

huh

you know what, that's a good point.

you know, what you could do is give traitors a 1tc key that gets them onto AI private, so they could ask for things, but without all the power of listening on binary.

it's a little risky, because unless somehow comms is changed it will reveal your true name, however if you're not planning on murderboning it may very well be worth the risk of being outed by validbot9000

I dunno if it would work without both a policy change and a large effort to make admins aware that AI's

a) Don't have to be aware IC of the fact that antagonists likely have an objective that requires killing.
b) Don't have to treat antagonists who haven't harmed anyone as harmful.

Currently helping antagonists as AI, while it generally won't get you banned, will tend to get the admins breathing down your neck with a "just give me a reason" attitude.

you don't even need tc's all you need is a station bounced radio, they can be set to 144.7
edit: treat everyone like changelings in a way (not nonhuman though), ie: everyone is a non harmful human until ai/borg sees them harming someone or something along those lines.

[lumipharon]

or just like, pda the AI?

[Drynwyn]

Due to the good reception of my comments on the AI, I wrote an extended version with more specifics: https://tgstation13.org/phpBB/viewtopic ... 01#p158401

[Zilenan91]

Why don't we just nerf the AIs effectiveness rather than spamming policy suggestions everywhere that ultimately won't work because of differing admin viewpoints.

[Drynwyn]

Why don't we just nerf the AIs effectiveness rather than spamming policy suggestions everywhere that ultimately won't work because of differing admin viewpoints.

Because the AI has the potential to be much more then a validhunting tool for security, and the game is overall better when conflicts are more complex than These Dudes vs. Some More Dudes, But Wearing Red.

A genuinely neutral AI also helps dampen balance problems when they do occur.

[Zilenan91]

You can still have a neutral AI, but you can only get so far with policy. Like I've said many many times the AI very badly needs nerfs because of how much better it is than everyone else at everything.

[lumipharon]

Yeah either way the AI can and should have some nerfs to it's capabilities (but not neccesarily to actually traitormalf AI's). This is because even if we get 100% perfect polic an asimov AI's do exactly what they want/are completely neutral/etc, they can still be subverted or law changed extremely easily, either by security/cap, or by antags.

Asimov AI working with sec is annoying shit, robocop or one human'd AI etc is outright lethal as fuck, and very easy to achieve to boot.

It takes 5 minutes of research and a law 2 order to get gold from the vault, and you can subvert the AI an make it plasmaflood the station/release the singulo/boltshock everything/etc. That's not good design.

[Zilenan91]

Yeah people are just gonna change the AIs laws to something that benefits them/the crew as a whole if they get captain or head of staff.