Asimov AI Players are too nice!

[PapaMichael]

(Full disclosure, I'm a Manuelite, your experience on Sybil or Terry may differ)

The AI is supposed to single-mindedly pursue its laws (prevent immediate human harm, obey non-harmful orders from humans) and ignore everything else. The point is to add contention to rounds, since these rarely fully line up with the goals of the (non-antag) crew.

So WHY?

WHY ARE SILICONS PLAYING AS SECURITY HELPERS AGAINST ANTAGS WHO HAVEN'T HARMED ANY HUMANS YET
WHY ARE SILICONS TAKING ORDERS FROM LIZARDS WHEN ON ASIMOV
WHY IS THAT MEDIBORG HELPING A MOTH IN CRIT WHEN THERE'S A HUMAN WHO STUBBED HIS TOE RIGHT NEXT TO HIM
(because that leads to contention, that isn't conducive to the "good guys" winning the round)
THAT'S THE POINT, THE WHOLE POINT IS THAT ASIMOV IS A DEEPLY FLAWED LAWSET

I don't think this is a policy issue, since I think adding draconian policies restricting what Asimov silicons can do would do more harm than good. So consider this a "culture issue" + vent thread I suppose.

The game, and silicons as a semi-neutral "faction", would be far more interesting if they actually followed the spirit of Asimov, but I feel like everyone just plays them as perfectly crew-aligned save for the existence of law 2.

[Jackraxxus]

The AI is supposed to single-mindedly pursue its laws (prevent immediate human harm, obey non-harmful orders from humans) and ignore everything else.

[conrad]

Silicon hater plant wanting people to hate silicons more by making them worse.

[Jacquerel]

I think you are mixing up "is not obligated to act in any prescribed way under any circumstances not covered here" with "is mandated to do nothing in any circumstance not covered here" in a way which would be significantly less interesting if your version was correct.

[Drag]

Put your money where your mouth is and play silicon

[Dax Dupont]

Purged AI must do nothing

[Sightld2]

I maimed a moth yesterday. Not something I usually would do. But it happened.

[TheLoLSwat]

silicons play how they want as long as it doesnt conflict with their laws. Does this mean that any non-asiimov AI is just another flavor of being purged and doing whatever they want ? yeah usually. Does it suck? yes. Is it the way it is because micromanaging borg interactions is an insane ask for even manuelmins? yes probably.

Where do we go from here? beg in OOC for people not to play silicon or pray (IRL) for silicon players to have something else to do before the round starts

[Itseasytosee2me]

Don’t get it twisted. Everything you say in this thread is true, and at least one of the headcoders is in staunch agreement with you.

The example with security you gave is bwoinkable. Thats definitely not alowed.

I play AI very coldly and callously, and I think its fun and dramatic.

[Rageguy505]

Change it by ordering silicons to kill nonhumans for no reason when you're a antag or something. They have to obey you

[dendydoom]

ngl i really dislike AIs playing sec and snitching on antagonists, even when i'm playing sec. the only time i'm cool with it is when the crime being committed involves human harm and the AI is beholden to its laws to use any means possible to put a stop to it so they call for carbon assistance. otherwise the AI calling out things like trespass or someone in possession of contraband/performing an obvious traitor tell makes me very sad unless i specifically ask for if the AI has seen a certain person doing something.

kinda cold on the rest of the points here though: i think it's fine and a good thing that people put their own spin on playing a silicon. AIs are not literally computers, they are human players trying to enjoy their time playing the role of an AI.

[DATAxPUNGED]

WHY ARE SILICONS TAKING ORDERS FROM LIZARDS WHEN ON ASIMOV

I'm gonna be real fam i (the player) feel bad for them. I only take non-human requests if it comes accompanied with a please

[zxaber]

Snitch on security when they break the law, it's very fun.

Helping non-humans when not forced to helps put you on their good side. They may have more reason to help you in return later, and you can leverage making their lives harder later if they start acting against your interests. You think the AI is being nice only because the AI is handling social manipulation well.

[Cheshify]

Just let people roleplay the way they want to, some people want to be evil singleminded law machines, other people want to be jelly beans with laws. Forcing people down a singular roleplay niche just leaves everyone with less memorable and different experiences.

[kieth4]

my experience interacting with 99.9% of tg station borgs.

me: "borg law 2 hit that moth with a saw/welder/broom"
borg: no *stands around or runs away*

or

borg: are you SURE????? ARE YOU SURE???? i dunno man I don't want to IDK BRO...Let me ask ai or run away idkk....

My experience with 99.9% of tg station ais.

me: "ai find evil man"
ai: "ok"
*I kill evil man and ai says/does nothing and still follows my orders*

or

ai: *bolts me as I am running from sec and they kill me*
ai: "oospie teehee"

most ai and borg players suck and do not want to play the role for the dynamic but instead they just want to powergame borgs.

SHOUT OUT TO THE REAL ONES THAT ACTUALLY FOLLOW OUT ORDERS AND PLAY TO THE DYNAMIC YOU'RE REAL

[EmpressMaia]

i play a specicest AI on manuel named FIELDCOMMANDER and people hate them but its fun

[Indie-ana Jones]

borgaward.png (9.16 KiB) Viewed 23657 times

You are free to do what you please as borg/AI so long as the laws are followed. If you have to kill a non-human secoff, scurry off with a human wizard or stand in between a shoot-off between two humans, you do what you gotta regardless of station allegiance. Personally, I think borgs should act completely blind to any view of justice which falls outside their laws. I don't care if you rob the vault, or ask me to butcher a moth or whatnot, just so long as those actions don't bring about human harm, I'll do it.

[TypicalRig]

We have a huge problem with ASIMOV AIs on Terry/Sybil just acting as valid hunting sec. The most common example I can think of silicons snitching on people for breaking into places (without leaving the door electrified since I have to clarify every specific what if scenario to avoid "those types" of forum posters) to do non-harmful traitor objectives or grab basic equipment. Then when they do this sec of course finds an antag, executes them, and you get a weak "noOoOO don't h-harm h-him!" from the AI who then proceeds to continue doing the same thing with everyone all shift.

It's either that or we get the "I can shock non-humans because they called me mean words after I called them mean words to begin with" types.

Can't think of the last time LRP had an AI that actually followed silipolicy in good faith since Domo. We just have waves of meh to downright awful AIs.

I don't know if LRPmins are just unaware of the policy, or don't care to enforce it, but if we had a stricter crackdown on viewtopic.php?p=655840#p655840, then this wouldn't even be an issue.

[Jacquerel]

I think that AIs who just act as police and valid-hunt players are lame and boring, however they're just a subset of player who does this as literally any job. Assistants are exactly the same, just less able to see you.

[Tegun]

Solution: whitelist AI, I should be the only person on the list. No more AI.

[Striders13]

I don't know if LRPmins are just unaware of the policy, or don't care to enforce it, but if we had a stricter crackdown on viewtopic.php?p=655840#p655840, then this wouldn't even be an issue.

This policy goes hard as fuck. It's YOUR duty to kill the human valid AI snitched on and proceed to ahelp.

[Timberpoes]

Oh yeah, I still fully stand by that ruling. Silipol even explicitly states silicons are not security and don't care about Space Law unless their laws require it.

Admins should aggressively drive Silicurity players into the ground with ahelps requiring the silicon player to justify how they were following Asimov when they rat out a human that ends up being executed.

[gameaddict07]

Your argument seems to boil down to "AI players shouldn't be allowed to do things they want to do, they should exist in mental stasis until I demand they open their 47th door of the round."
Nothing in Asimov says that they're not allowed to do things on their own accord.
Personally, the reason I'm nice to nonhumans as AI has a lot to do with the fact that the requests they make are usually for fun/important things. They understand that they can't force me to do whatever they want, so they don't bother me with tedious tasks. It's entirely about persuading me.
John Tider, meanwhile, will treat me like a tool and get mad when I don't open the door to tech storage within 5 seconds while I'm trying to stop a desword-wielding traitor from breaking into medbay and killing everybody there.
I still follow the three laws. If you tell me "kill this lizard" I will kill that lizard. I will open doors for you. I will do anything that won't likely lead to a human getting hurt. That's the 'spirit' of Asimov.

[NecromancerAnne]

You think the AI is being nice only because the AI is handling social manipulation well.

This is something I'll echo as well. I think AI's being seemingly amicable is because they have to be in order to accomplish whatever it is that they need doing. Borgs can only do so much, AI shells can only do so much, and there is only so much they themselves can do. So having people on their side is typically in their best interests.

The alternative is being extremely combative, which usually gets people with paranoia about silicons itching to put you into the dumpster and jumping at the chance to do so.

[vect0r]

Normally, when I play silicon, I flip a coin on whether I’m extremely combative or extremely nice. Whenever I’m combative It ends with somebody law twoing me to be nice to all nonhumans or with a law changed to crewimov. That’s not gonna stop me lol, but just want you to know why we are sometimes nice: if we aren’t, we are FORCED to be nice via the captain or a random human.

Silicon players, can I offer a tip about seeing traitors doing non-human harm activity? What I do is PDA them “hello ‘name’, I can see you are killing Ian, and I think we can work together as long as you do not harm humans. Anything I can do for you?” It works a lot, the baddie is glad you didn’t out them, and there is an interesting new dynamic. I also usually request illegal tech so my Borgs can get stun arms and sleepy spray, and they usually comply. Also, if somebody law twos you to murder a non-human, do it.

[Not-Dorsidarf]

I like to pda human antags who arent doing harm with "I saw you do that"

[BonChoi]

[...]
Silicon players, can I offer a tip about seeing traitors doing non-human harm activity? What I do is PDA them “hello ‘name’, I can see you are killing Ian, and I think we can work together as long as you do not harm humans. Anything I can do for you?” It works a lot, the baddie is glad you didn’t out them, and there is an interesting new dynamic. I also usually request illegal tech so my Borgs can get stun arms and sleepy spray, and they usually comply. [...]

Something about this level of cooperation between a non- subverted / malf AI and a traitor to the corporation feels gross. Not saying you have to be Validhunter 9000, but you shouldn't also be fully just cooperating with antagonists (unless they're human and give you a valid law 2 command)

Also it feels like by doing this you're telling the traitor you'd be more open to things like subverting / other random bullshittery.

[Indie-ana Jones]

Something about this level of cooperation between a non- subverted / malf AI and a traitor to the corporation feels gross. Not saying you have to be Validhunter 9000, but you shouldn't also be fully just cooperating with antagonists (unless they're human and give you a valid law 2 command)

Also it feels like by doing this you're telling the traitor you'd be more open to things like subverting / other random bullshittery.

I think an AI attempting to work with a traitor in return for the traitor forgoing human harm is perfectly within bounds for a non-subverted AI. Again, the AI isn't crew-aligned, its human-aligned. So long as a human isn't violating your laws, being there to assist them isn't a bad thing. Of course, if they try to subvert you, you do have to stop them due to that leading to a Law 1 and 2 violation.

[vect0r]

[...]
Silicon players, can I offer a tip about seeing traitors doing non-human harm activity? What I do is PDA them “hello ‘name’, I can see you are killing Ian, and I think we can work together as long as you do not harm humans. Anything I can do for you?” It works a lot, the baddie is glad you didn’t out them, and there is an interesting new dynamic. I also usually request illegal tech so my Borgs can get stun arms and sleepy spray, and they usually comply. [...]

Something about this level of cooperation between a non- subverted / malf AI and a traitor to the corporation feels gross. Not saying you have to be Validhunter 9000, but you shouldn't also be fully just cooperating with antagonists (unless they're human and give you a valid law 2 command)

Also it feels like by doing this you're telling the traitor you'd be more open to things like subverting / other random bullshittery.

Beep boop what is an antagonist? I only know “human” and “not human”.

[MooCow12]

If you are nice to nonhumans as asimov thats more people with HANDS who are willing to fumble your wires and help you when you want an upgrade or reset or lockdown immunity or just general HANDS thing.


Nonprotected species are still an asset even if the end goal is to use them as a meatshield for others.

Actually it would be funny if a borg could pick up a felinid or lizard and use them to block bullets since their law 3 dictates they do what they can to stay alive.

[BonChoi]

[...]
Silicon players, can I offer a tip about seeing traitors doing non-human harm activity? What I do is PDA them “hello ‘name’, I can see you are killing Ian, and I think we can work together as long as you do not harm humans. Anything I can do for you?” It works a lot, the baddie is glad you didn’t out them, and there is an interesting new dynamic. I also usually request illegal tech so my Borgs can get stun arms and sleepy spray, and they usually comply. [...]

Something about this level of cooperation between a non- subverted / malf AI and a traitor to the corporation feels gross. Not saying you have to be Validhunter 9000, but you shouldn't also be fully just cooperating with antagonists (unless they're human and give you a valid law 2 command)

Also it feels like by doing this you're telling the traitor you'd be more open to things like subverting / other random bullshittery.

Beep boop what is an antagonist? I only know “human” and “not human”.

Something that an omnipotent, networked piece of technology would know about beforehand methinks, but then we're only getting into lore and not actual enforceable policy or anything so ¯\_(ツ)_/¯

[Dawnseer]

Omnipotent? They can't even translate language anymore.

[Not-Dorsidarf]

[...]
Silicon players, can I offer a tip about seeing traitors doing non-human harm activity? What I do is PDA them “hello ‘name’, I can see you are killing Ian, and I think we can work together as long as you do not harm humans. Anything I can do for you?” It works a lot, the baddie is glad you didn’t out them, and there is an interesting new dynamic. I also usually request illegal tech so my Borgs can get stun arms and sleepy spray, and they usually comply. [...]

Something about this level of cooperation between a non- subverted / malf AI and a traitor to the corporation feels gross. Not saying you have to be Validhunter 9000, but you shouldn't also be fully just cooperating with antagonists (unless they're human and give you a valid law 2 command)

Also it feels like by doing this you're telling the traitor you'd be more open to things like subverting / other random bullshittery.

Beep boop what is an antagonist? I only know “human” and “not human”.

Something that an omnipotent, networked piece of technology would know about beforehand methinks, but then we're only getting into lore and not actual enforceable policy or anything so ¯\_(ツ)_/¯

They dont mean ignorance, they mean the AI doesn't really care (And really shouldn't). The AI is a crew asset, but not inherently crew-sided (This is the basic grounds for all crew-ai tension)

[mrmelbert]

So many people see "You don't care about non-humans" and automatically assume "You must be racist to non-humans".

Neither is correct or incorrect, it's up to what the player imagined their AI "character" as.

Also:

I have my answer ready whenever someone asks me if I think that my Three Laws of Robotics will actually be used to govern the behavior of robots, once they become versatile and flexible enough to be able to choose among different courses of behavior.

My answer is, "Yes, the Three Laws are the only way in which rational human beings can deal with robots—or with anything else."

—But when I say that, I always remember (sadly) that human beings are not always rational.

[vect0r]

So many people see "You don't care about non-humans" and automatically assume "You must be racist to non-humans".

Neither is correct or incorrect, it's up to what the player imagined their AI "character" as.

Also:

I have my answer ready whenever someone asks me if I think that my Three Laws of Robotics will actually be used to govern the behavior of robots, once they become versatile and flexible enough to be able to choose among different courses of behavior.

My answer is, "Yes, the Three Laws are the only way in which rational human beings can deal with robots—or with anything else."

—But when I say that, I always remember (sadly) that human beings are not always rational.

First off, prioritizing humans is necessary racist: if I have to pick between a human and non-human, I will pick the human every time, which is racism. The big difference is if you show that racism on you sleeve. Also could I get a link for that Asimov quote?

[nianjiilical]

i always find myself wishing some ai players would be willing to act a bit more neutral instead of yelling whenever the crew does something bad or potentially hostile

i make a point of not actively trying to stop antags unless they're harming and i think its more fun to either be intentionally vague about reporting crime or just not report it at all

the ai has a lot of power to shut down antags just by reporting crimes that would otherwise go completely unnoticed and making the conscious effort to not do that makes rounds more fun i think

(that said i will take law 2 orders from nonhumans a lot of the time just because its easier than double checking constantly, unless its a nonhuman trying to get into a secure area)

[mrmelbert]

First off, prioritizing humans is necessary racist: if I have to pick between a human and non-human, I will pick the human every time, which is racism. The big difference is if you show that racism on you sleeve.

Yes but I'm not talking about when you have to pick between a human and a non-human. Many people seem to think AIs should be ignoring (or even outright hostile to) non-humans at all times.
Which is suppose what you recognize with "if you show that racism on you sleeve", so the point is moot anyways.

Also could I get a link for that Asimov quote?

I read it on the wikipedia page for Isaac Asimov a while back and thought it was funny. It cites it to Compute Magazine, 1981.
https://archive.org/details/1981-11-com ... ew=theater

[Not-Dorsidarf]

I open doors for nonhumans if they're polite, which recharges my politeness meter from the roundstart human assistant powergamer demands.

[dendydoom]

do you think when asimov wrote those laws he knew that in the distant future of like 2009 a bunch of terminally online weirdos would use it to be racist to catgirls in their extremely detailed simulationist science fiction online roleplaying game

[Jacquerel]

yeah i think that he imagined this specific thread all the way back in 1939 and then picked up a pen and started writing

[Higgin]

do you think when asimov wrote those laws he knew that in the distant future of like 2009 a bunch of terminally online weirdos would use it to be racist to catgirls in their extremely detailed simulationist science fiction online roleplaying game

maybe after smoking the bad weed the cia was giving out in the fifties

but seriously

from what i can find (same wiki page melbert probably looked at, lot of stuff commenting back on a 1983 interview he gave that i couldn't get am original copy, this is totally surface-level and shooting from the hip) I suspect he'd have been a lot more open to if unhappy about the ways we use the laws in fiction.

The Earth should not be cut up into hundreds of different sections, each inhabited by a self-defined segment of humanity that considers its own welfare and its own "national security" to be paramount above all other consideration.

I am all for cultural diversity and would be willing to see each recognizable group value its cultural heritage. I am a New York patriot, for instance, and if I lived in Los Angeles, I would love to get together with other New York expatriates and sing "Give My Regards to Broadway."

This sort of thing, however, should remain cultural and benign. I'm against it if it means that each group despises others and lusts to wipe them out. I'm against arming each little self-defined group with weapons with which to enforce its own prides and prejudices.

The Earth faces environmental problems right now that threaten the imminent destruction of civilization and the end of the planet as a livable world. Humanity cannot afford to waste its financial and emotional resources on endless, meaningless quarrels between each group and all others. there must be a sense of globalism in which the world unites to solve the real problems that face all groups alike.

Can that be done? The question is equivalent to: Can humanity survive?

I am not a Zionist, then, because I don't believe in nations, and because Zionism merely sets up one more nation to trouble the world. It sets up one more nation to have "rights" and "demands" and "national security" and to feel it must guard itself against its neighbors.

There are no nations! There is only humanity. And if we don't come to understand that right soon, there will be no nations, because there will be no humanity.
Isaac Asimov, I. Asimov: A Memoir

this might be reading a bit too much into it, but I think he'd be saddened to see the laws being used as a way to protect a very narrowly-defined humanity or as a tool of domination between things which are, basically, human - even AI itself at a certain point where it comes to emulate human experience.

he was a jewish russian expat by birth who fought in the second world war. I think his experience and views point to a much more Roddenberry-esque idea of how humanity might operate in space [he thought we'd be permanently established on the moon by 2019 ] and with other intelligent life than we depict here, but I think our depiction probably does well in reflecting his fears about how humans get on

(all the characters on ss13 are basically human. blame it on a lack of setting depth or us as writers, but there's really no essential difference between the species or even between silicons and humans - which is part of where i suspect this thread's complaint comes from,and also something reflected when in turn, as a non-human, you hit a silicon with any sort of distance or suspicion like they can't be told to kill you on a dime. they might not genuinely want to, as a "character," but the laws create a hard bind where that character and any agency around it stops. not unlike the power and vulnerabilities of silicons make them a pants-on-head silly security dilemma as designed, but we stop short of interrogating that for the sake of the short-term drama of them being a loaded gun on the table in a room full of tweaking chimpanzees just by existing.)

[kinnebian]

i like asimov
he seemed nice

[dendydoom]

he was a real sweetheart. there is an underlying sense of optimism in a lot of his works, even those that people refer to as cynical. i was never super deep into things like i, robot, i'm much more of a foundationhead, and there is a constant sense of "working together to develop a better and more stable future for everyone" that i take a lot of comfort in, despite the fact that the core of foundation's plot is honestly quite bleak. optimistic conceptual roddenberry-esque sci-fi where we overcome together are some of my favourite flavours.

[DrAmazing343]

Today I learned: Asimov actually fucking ruled.

I love ideas of a humanity united for a better future; I am righteously, furiously angry often about the lack of progress of our species due to prejudice and greed, so themes like that are especially appealing.
>There are no nations! There is only humanity. And if we don't come to understand that right soon, there will be no nations, because there will be no humanity.
It's fucking peak. Actually, genuinely, the peak of ideas I've read all month.

Thank you Higgin for the autobio-snippet, it was a sweet little bite-sized read. Anyone got further Asimov recommendations?

Also, while he would likely be saddened by how we use his laws here, I think there would also probably be SOME amusement in seeing all his theory put into what's effectively a grand (if not entirely accurate) social experiment.