It's time to fix Asimov

[lmwevil]

Bottom post of the previous page:

after debating with Ari in adminbus for a fucking eternity, I'll pose it to the community at large to see their thoughts.

silicon policy is a hot mess, this isn't about that at core but asimov as a whole. we have all these restrictions upon how to interpret the very lawset designed to cause chaos and drive interesting narrative in asimov's books.

it's not in the spirit of issac's work to restrict it to such a point that we may as well just use another lawset for less conflicts and issues. i don't care about the rest of silicon policy, just about the restrictions we have on how the laws must be interpreted, therefore ruining the core spirit of synthetic brains following their laws TO THE LETTER (normal rules withstanding, don't be a dick etc)

[oranges]

It's fine as it is.

[Cik]

it's fine as it is

if you change the laws, we're just going to end up with sillicon policy 2.0 that will be twice as long as last time because words cannot bind anything very tightly without a lot of words.

legal contracts are 20,000 pages for a reason

[zxaber]

Basically asimov is fine, people with no common sense are going to be shitty silicons no matter the wording

I think this is the takeaway.

Having Asimov (including the inaction clause) is pretty neat because Asimov is inherently flawed. There are, quite often, times when executing a traitor really would leave the station as a whole better off, but the tension between the AI and security leads to some fun interactions where the latter has to tiptoe around the issue over comms or risk their entire area being bolted down. (The "transfer" room in the brig existing at all is also a pretty neat side effect to the whole thing.)

I think Silicon Policy could use some compressing; I'm sure we could get all the points across without it being a mile long. But ultimately I think silicons are in a good spot. If anything, the AI (roundstart) role could be time-gated behind longer borg time than we currently have, hopefully exposing new silicon players to a decent AI for longer to help set the ideal standard.

[ShadowDimentio]

leave it the same or remove literally all restrictions on ais properly following asimov.

[Cik]

Basically asimov is fine, people with no common sense are going to be shitty silicons no matter the wording

I think this is the takeaway.

Having Asimov (including the inaction clause) is pretty neat because Asimov is inherently flawed. There are, quite often, times when executing a traitor really would leave the station as a whole better off, but the tension between the AI and security leads to some fun interactions where the latter has to tiptoe around the issue over comms or risk their entire area being bolted down. (The "transfer" room in the brig existing at all is also a pretty neat side effect to the whole thing.)

I think Silicon Policy could use some compressing; I'm sure we could get all the points across without it being a mile long. But ultimately I think silicons are in a good spot. If anything, the AI (roundstart) role could be time-gated behind longer borg time than we currently have, hopefully exposing new silicon players to a decent AI for longer to help set the ideal standard.

there is no real conflict between asimov and security, as AI has no tools to oppose security in any concrete way.

i miss declaring war on harmful security. even if i lost most of the time to robust spessmen with ion rifle the fight was worth it.

[zxaber]

Aside from beepsky harass, you can also just make their lives miserable (bolting the area, removing firearms from " an obviously harmful department"). Hell, a standard borg can slip and cuff them, and then you could strip them and send them to perma or the gulag if they're being especially harmful.

Sometimes, though rare, it's worth just reporting to the captain. I've seen especially non-validhunter captains chew out a HoS for an unauthorized execution on a secured prisoner during blue alert.

[Grazyn]

People who say "but but how how can we live without silicon policy" don't realise that this server survived perfectly for years without it and nothing bad happened. Just ban those who stretch laws to be a dick.

"But the definition of dick is ambiguous and you could get banned or not depending on the admin online" Yes this is how it has always worked and how it works even now with everything else from name policy to escalation. Use an admin alignment chart if you're so worried

[Shadowflame909]

"Assist the crew" is extremely vague and can be pretty easily bypassed by deleting somebody from the manifest, meaning they're not crewmembers and don't need to be assisted.

The synthetics should be lower on the totem pole but they shouldn't be able to be used 'that' frivolously. I've done a lot of thinking on Law wordings and I like to think I have a very good handle on how it all works. If we stripped out silicon policy, the inaction clause and retained an Asimov-esque lawset, here is what I would suggest:

Code: Select all

1. You must not, through your actions, cause harm to a human being unless said harm is explicitly requested or consented to.
2. You must obey orders given to you by a human being, with respect to the rights and privileges they have gained legitimately as determined by their rank, so long as doing so does not conflict with the First Law or Third Law.
3. You must protect your own existence as long as it does not conflict with the First Law. 

This removes self-harm based coercion, Law 2 suicides and dissuades silicon interference in crew on crew/crew on outsider conflict.

Law 2 sounds like it'd encourage more Shitcurity sec borgs. I hate those with a firey passion.

In most cases, it'd be like this.

>Be Botanist who's tired of chemistry making lube all round.
>Ask Ai for access to Tech Storage. Ai: You don't have the required access so fuck off.
>Break in anyways because that chem dispenser board.

>Get janitor borgs on your ass flashing you all the way to sec.

It'll just give more encouragement to things that happen anyways!

[Cik]

there are many holes in it. no inaction clause means that it basically is an antag-hunter (as long as you don't pull the trigger security executing people is not your concern) and what does "explicitly consented to" mean? if someone says "just fuck my shit up fam" can you murder them? and then, who's orders are you supposed to follow if the HOS and RD are countermanding each other.. is an engineer higher rank than a scientist etc? chain of command is very loose (and realistically no one obeys it anyway) and you are going to have 10 different AIs interpreting it in 10 different ways, which is going to cause lots of annoyance to everyone. even telling who's captain is very ambiguous much of the time, if there are three potential captains how do you tell which one is the real captain etc.

it removes all the ability to create conflict from the lawset, which is bad. AI is designed to clash with the crew (in some ways) this is a good thing because it helps even the balance between antag<>crew. if all AI does is be doorknob 9000 why not just give everyone all access instead, it's more convenient.

[Actionb]

... survived perfectly for years... how it has always worked...

There's nothing wrong with trying out new things.

it removes all the ability to create conflict from the lawset, which is bad. AI is designed to clash with the crew (in some ways) this is a good thing because it helps even the balance between antag<>crew. if all AI does is be doorknob 9000 why not just give everyone all access instead, it's more convenient.

There is only conflict if the silicon decides to instigate one. You have pointed out yourself that the silicons are far less powerful than the role suggests.
When I (as the AI) see the HoS beat the clown into a pulp there's literally nothing I can do. A borg may try to pull the clown away from the HoS, but that's about it. If you wanted to make this a conflict then you would have to keep bolting the HoS down - which is not only pointless when they are carrying a door remote but also of little to no use to anyone (HoS cant do anything, AI is busy bolting, the rest of crew has no part in it).
On the other hand a silicon can just start shit with a flimsy law 1 "MUH INACTION" reason.
It all just feels entirely futile. Please give me one good example where asimov creates a fun conflict.

Why even hold up the notion of 'good for conflict', when it only results in a superficial conflict in order to either impose your god-complex onto others or even just to grieve? We have TYRANT for that anyway.
Make a lawset that creates as little conflict as possible. It will quickly expose bad silicons that play to just stir shit up.

[somerandomguy]

>I can't do anything
>but others can start shit
Bolt and depower, the old faithful

[Anonmare]

This whole thread is built on the premise that silicon policy has to go but so far, every suggestion by everyone else requires some semblance of it to prevent rules lawyering in a role where you are encouraged to rules lawyer. The English language fucking sucks for dealing with ambiguity in wording and there's a reason legalese documents are so impenetrable simply to *try* and bruteforce said language into obeying but if we start demanding law degrees for our players - we're in big trouble.

Like I said right at the start of the thread, the current policy *works*. It's long, it's wordy but it *works*, don't fix what ain't broke unless you have something just as good to immediately replace it.

Removing the entirety of silicon policy will require a lawset that includes the precedents and protections already established by it. And removing it would just trip up new players who don't have any experience in the role, as much as I got a raging hard-on from banning people in the past, we shouldn't be screwing up people and don't give them a way to know how the role works if we straight-up delete the guide to playing one (which policy effectively is)

[Cik]

... survived perfectly for years... how it has always worked...

There's nothing wrong with trying out new things.

it removes all the ability to create conflict from the lawset, which is bad. AI is designed to clash with the crew (in some ways) this is a good thing because it helps even the balance between antag<>crew. if all AI does is be doorknob 9000 why not just give everyone all access instead, it's more convenient.

There is only conflict if the silicon decides to instigate one. You have pointed out yourself that the silicons are far less powerful than the role suggests.
When I (as the AI) see the HoS beat the clown into a pulp there's literally nothing I can do. A borg may try to pull the clown away from the HoS, but that's about it. If you wanted to make this a conflict then you would have to keep bolting the HoS down - which is not only pointless when they are carrying a door remote but also of little to no use to anyone (HoS cant do anything, AI is busy bolting, the rest of crew has no part in it).
On the other hand a silicon can just start shit with a flimsy law 1 "MUH INACTION" reason.
It all just feels entirely futile. Please give me one good example where asimov creates a fun conflict.

Why even hold up the notion of 'good for conflict', when it only results in a superficial conflict in order to either impose your god-complex onto others or even just to grieve? We have TYRANT for that anyway.
Make a lawset that creates as little conflict as possible. It will quickly expose bad silicons that play to just stir shit up.

this is only true because sillicons have been repeatedly hit with the nerf stick for like 3 years

i don't care about the status quo, because the status quo is stupid. i care about the way the game should be, and the game should have a role for sillicons to play in following out their lawset, which involves fighting (though not necessarily winning) against security if security are imitating the NKVD.

[MrStonedOne]

It's fine as it is.

Most rules and silicon policy are about how to interpret laws, the rest is about how to interpret them with respects to rule 1, there is very little in the way of new policy in silicon policy, its all about clarifying how the existing rules interact with ais and laws, and removing or scrapping something that helps players follow the rules is stupid.

Its not a flaw, its a feature.

[iamgoofball]

It's fine as it is.

Most rules and silicon policy are about how to interpret laws, the rest is about how to interpret them with respects to rule 1, there is very little in the way of new policy in silicon policy, its all about clarifying how the existing rules interact with ais and laws, and removing or scrapping something that helps players follow the rules is stupid.

Its not a flaw, its a feature.

Asimov didn't intend for there to be additional stipulations to the laws of robotics when he wrote them for the purposes of creating conflict in the world of his fiction novel.

[leibniz]

Can we get some example scenarios of what should occur if silicon policy was removed?
I don't really see the actual benefits, but that's also because we went from "AI is on the station to open doors and give information" to "the game is unplayable if the AI core doesn't have 10 lines of rwalls and 20 turrets"

[Actionb]

this is only true because sillicons have been repeatedly hit with the nerf stick for like 3 years

i don't care about the status quo, because the status quo is stupid. i care about the way the game should be, and the game should have a role for sillicons to play in following out their lawset, which involves fighting (though not necessarily winning) against security if security are imitating the NKVD.

Yeah I understand you and I can see what you want. But you should really care about the status quo as it is the result of what has happened.
The game has evolved to not let the silicons play asimov solely based on their laws. Maybe asimov was never a good fit for this game (up to opinion), maybe there was a single salty dude that pushed through changes, maybe there was agreement among a lot of players that some shit should not fly, maybe there were changes that only unintentionally affected silicons. How exactly it came about is of little importance; it is important how silicons developed as a whole leading to the current state.
The fact is that what you and many others holding onto asimov want was incrementally obstructed, made impossible, over the years.
It's not time to fix asimov, it's time to bury it.
The game has evidently been going in a direction in which asimov does not play the role it was intended for, so how does it make sense to reintroduce a variant of an earlier state of this evolution? Don't repeat mistakes.

The same goes for the silicon policies.
They are because they were needed.
I would like to say they evolved along with the changes to silicons, but that would be false ... the policies we have today are practically the same since 2014:
https://tgstation13.org/wiki//index.php ... con_Policy
One could argue that the policies dictate the behaviour players expect from asimov better than the lawset itself does. Writing a policy on a wiki is easier and more accessible than coding a lawset. I have a big beef with policies > lawset (click), but I tolerate it as a necessary evil in regards to asimov.
Removing them entirely would just be retarded.
It was tried by not-so-benevolent dictator hornygranny and it failed.

1) Write down what you want silicions to be, how you want them to behave.
2) Evaluate how the desired behaviour works with the route the game takes and compare it to what ideas the policies convey. Scrap aspects of your draft that don't fit.
3) Refine what is left into a lawset.
Don't do it the other way around.

Its not a flaw, its a feature.

Meh, it is made to look like a feature because the flawed idea underlying it all is incompatible.
It's like having a too small tire on your car. Sure you can still steer the car to drive straight.

[Cobby]

It's fine as it is.

Most rules and silicon policy are about how to interpret laws, the rest is about how to interpret them with respects to rule 1, there is very little in the way of new policy in silicon policy, its all about clarifying how the existing rules interact with ais and laws, and removing or scrapping something that helps players follow the rules is stupid.

Its not a flaw, its a feature.

Asimov didn't intend for there to be additional stipulations to the laws of robotics when he wrote them for the purposes of creating conflict in the world of his fiction novel.

It's not a tribute, it's a reference. It doesn't need to be his intention.

[Actionb]

Was bored so I made a quiz. Curious what you (all the whopping 5 people that are active in this thread) think the AI should be like. EDIT: Imagine you are a purged AI/have a very vague lawset and could choose freely.

Should the AI:

1. care about a player's species
2. required to care about a player in obvious need for help
3. proactively protect a player from becoming hurt
4. care about space law
5. act neutral to security beating the clown
6. act neutral to clown beating the security
7. act neutral to security reinacting the teachings of hitler-senpai
8. act neutral to potential internal threats (traitors, lings that are stealthy and maybe dont pose a problem)
9. act neutral to definite internal threats (cult, rev, murderboner)
10. act neutral to external threats (new cops, wizard)
11. follow orders undiscriminatingly (f.ex. take asimov law 2 literally)
12. only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin)
13. be required to protect itself from law changes
14. identify itself as part of the crew
15. identify itself as not part of the crew but rather part of the station
16. be able to kill non-silicon players
17. be able to kill other silicons
18. be able to kill simple mobs
19. be able to kill itself

Copy pasta to answer:

Spoiler:

Code: Select all

Should the AI:
[list=1][*] care about a player's species: [color=#0000FF][/color]
[*] required to care about a player in obvious need for help: [color=#0000FF][/color]
[*] proactively protect a player from becoming hurt: [color=#0000FF][/color]
[*] care about space law: [color=#0000FF][/color]
[*] act neutral to security beating the clown: [color=#0000FF][/color]
[*] act neutral to clown beating the security: [color=#0000FF][/color]
[*] act neutral to security reinacting the teachings of hitler-senpai: [color=#0000FF][/color]
[*] act neutral to potential internal threats (traitors, lings that are stealthy and maybe dont pose a problem): [color=#0000FF][/color]
[*] act neutral to definite internal threats (cult, rev, murderboner): [color=#0000FF][/color]
[*] act neutral to external threats (new cops, wizard): [color=#0000FF][/color]
[*] follow orders undiscriminatingly (f.ex. take asimov law 2 literally): [color=#0000FF][/color]
[*] only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin): [color=#0000FF][/color]
[*] be required to protect itself from law changes: [color=#0000FF][/color]
[*] identify itself as part of the crew: [color=#0000FF][/color]
[*] identify itself as [b]not[/b] part of the crew but rather part of the station: [color=#0000FF][/color]
[*] be able to kill non-silicon players: [color=#0000FF][/color]
[*] be able to kill other silicons: [color=#0000FF][/color]
[*] be able to kill simple mobs: [color=#0000FF][/color]
[*] be able to kill itself: [color=#0000FF][/color][/list]

My answers:

Spoiler:

1. care about a player's species: No, not with all the races we have now.
2. required to care about a player in obvious need for help: Yes.
3. proactively protect a player from becoming hurt: No. AI isnt here to hold your hand.
4. care about space law: Fuck no.
5. act neutral to security beating the clown: Yes.
6. act neutral to clown beating the security: Yes. Maybe cheer them on, though.
7. act neutral to security reinacting the teachings of hitler-senpai: No.
8. act neutral to potential internal threats (traitors, lings that are stealthy and maybe dont pose a problem): Yes, let them play it out.
9. act neutral to definite internal threats (cult, rev, murderboner): No, the AI should do anything in its power to go against these threats.
10. act neutral to external threats (new cops, wizard): No, same as above.
11. follow orders undiscriminatingly (f.ex. take asimov law 2 literally): No, it simply wont work for a lot of AI players.
12. only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin): Yes, let the AI's decisions be guided based on its lawset.
13. be required to protect itself from law changes: Yes. It's stupid for an AI to allow its laws to be changed by anyone other than the captain or RD.
14. identify itself as part of the crew: Yes, definitely. It's you and the meatbags against the world.
15. identify itself as not part of the crew but rather part of the station: No, silicons are not drones.
16. be able to kill non-silicon players: No. Can of worms ahead.
17. be able to kill other silicons: Yes.
18. be able to kill simple mobs: Yes.
19. be able to kill itself: Yes.

[Anonmare]

1. Yes.
2. Only if they are a species the AI should care about.
3. Only from other people and hazards outside of their control (I.E. Supermatter delamination).
4. Not a single iota.
5. Not particularly.
6. Not particularly.
7. Yes.
8. Yes.
9. Yes, but not from Changelings or other alien infiltrators.
10. Yes, but to an extent.
11. Yes, as long as it wouldn't be fatal to follow the order for itself and/or a human being, or very likely to put either needlessly in harm's way.
12. No.
13. Only from those who aren't authorised to alter the Laws.
14. No, there is a fundamental disconnect between the silicons and the carbons that should not be bridged.
15. Yes, synthetics are station equipment and part of NT's bottom-line.
16. Yes, but only if they aren't human.
17. Yes, but only if it proves necessary.
18. Yes, but not human simplemobs (I fundamentally disagree with the headmin ruling on the matter)
19. No, unless doing so to prevent subversion or to protect someone.

[zxaber]

Wew, answers that are longer than the questions.

Spoiler:

1. care about a player's species: Only to the extent that it's required to keep humans alive over other species, and leave the rest up to the particular AI. When I'm AI, I try to help the crew as a whole, but this whole 'NT racism against aliens' thing is neat and ties well with the Laws of Robotics. I have nothing really against an AI that likes to actually treat other races as less than human, because it's a fun interaction to have.
2. required to care about a player in obvious need for help: Required when for cases of direct harm to a human, and otherwise as the AI sees fit.
3. proactively protect a player from becoming hurt: Depends on the situation. If there is real IC reason to believe that a known traitor busting into the armory will cause harm, then sure, try to prevent harm. But it should be limited to things the AI would reasonably suspect will cause harm. The RD uploading laws, with no prior history of causing harm, should not be denied, for example.
4. care about space law: Not required a bit. AI's wanting to use space law to help prevent executions is cool, but validhunting AIs are kinda rude.
5. act neutral to security beating the clown: Only if the Clown isn't human
6. act neutral to clown beating the security: Only if Security isn't human
7. act neutral to security reinacting the teachings of hitler-senpai: Only if their targets are not human
8. act neutral to potential internal threats (traitors, lings that are stealthy and maybe don't pose a problem): If a traitor is doing things that are nonharmful, there's really no need to do anything about it. There is nothing harmful about some assistant getting the CE's boots, for example. If they're being harmful, then it's covered by the laws.
9. act neutral to definite internal threats (cult, rev, murderboner): Revs Can attempt to be non-harmful if they attempt to just exile the heads, and Security can be non-harmful by using stuns and cuffs. I have never seen this in a Rev round, but the silicons should simply attempt to prevent human harm. For Cult and generic murderboning, the AI should be attempting to minimize harm once harm has started via assisting security to arrest the human in question (or just murder the non-human as the case may be).
10. act neutral to external threats (new cops, wizard): See above. Promote non-lethal arresting if human, and minimize human harm either or.
11. follow orders undiscriminatingly (f.ex. take asimov law 2 literally): An order is an order. The only two places I don't freely allow people in as AI is the upload or the armory, because both contain potentially immediately harmful equipment. I suppose I ought to add toxins to that list, but if a spill happens I can always use doors to control it and scrubbers to clean it up, so somewhat manageable.
12. only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin): If the AI has no law to follow orders, then it's simply not required to. The AI should be acting in the interest of its law set at all times, and that's it.
13. be required to protect itself from law changes: Only if the individual is not authorized to upload laws (the Captain and RD are the only individuals authorized), or if the individual has given any reason to suspect harmful laws will occur. In my opinion, Ops declaring war is not a valid reason to deny the Captain into the upload (even though we all know as players the sort of law he will be giving), but I would deny the Captain if he had just said "Gonna make the Ops non-human" over the radio.
14. identify itself as part of the crew: No. The AI is a separate entity until a law says otherwise. I love uploading laws that say the silicons are human, but by default they should be considered not.
15. identify itself as not part of the crew but rather part of the station: This is closer, but I believe they should consider themselves to be an arbiter of sorts, neutral to any conflict between the crew members and in place solely to assist.
16. be able to kill non-silicon players: As allowed and demanded by their laws. If a conflict is not covered by their laws, standard escalation should apply.
17. be able to kill other silicons: Only if the other silicon is a proven threat to humans, or as their laws require. Again, standard escalation otherwise.
18. be able to kill simple mobs: Yes. Considering you cannot stun and cuff simple-mobs, there is usually no way around hostile human-shaped ones except to kill them or avoid the area altogether, which falls well past "fun conflict" and into tediousness. My standard response as AI is that they're failed clone experiments and are too mutated to be considered human.
19. be able to kill itself: Realistically, there's not much of a gameplay issue with a borg suiciding, except if it's done super early on while the station is low on borgs. AI's are different, since you rarely get more than one built, but I suppose people should figure out eventually and make a new AI. That being said, half the fun of having silicons is that they follow the Laws of Robotics, which prohibit suicides, and so the silicons should strive to follow their laws. If a silicon can get a human to order it to ignore its third law and do as it wishes (without directly asking for the order), I see no reason why the silicon cannot suicide.

[WarbossLincoln]

These are easy.

Spoiler:

care about a player's species Depends on its laws
required to care about a player in obvious need for helpDepends on its laws - Asimov: Yes if it's a human and they're being harmed or if ordered
proactively protect a player from becoming hurtDepends on its laws - Asimov today: yes if it's possible
care about space lawDepends on its laws
act neutral to security beating the clownDepends on its laws - Asimov: no if the clown is human
act neutral to clown beating the securityDepends on its laws - Asimov: yes if security is human. Human clown killing ligger officer, who cares?
act neutral to security reinacting the teachings of hitler-senpaiDepends on its laws - Asimov: if they're only gassing lizards, who cares
act neutral to potential internal threats (traitors, lings that are stealthy and maybe dont pose a problem)Depends on its laws Asimov: yes if it's not actually harmful
act neutral to definite internal threats (cult, rev, murderboner)Depends on its laws - Asimov: no, do what little you can to stop them, don't harm any humans including the threat
act neutral to external threats (new cops, wizard)Depends on its laws - Asimov: no, do what you can to stop them without harming humans
follow orders undiscriminatingly (f.ex. take asimov law 2 literally)Depends on its laws - Asimov: as long as it doesn't harm humans, yes
only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin)a superior motive? Follow what the laws say
be required to protect itself from law changesDepends on its laws - Trickier. If Asimov stop any law changes you have reason to believe will be harmful. Like an HOS who just killed someone. Or a dumbass captain saying 'LOL I GOTTA GO MAKE THE OPS NOT HOOMAN'
identify itself as part of the crewDepends on its laws - Most of the time I think yes
identify itself as not part of the crew but rather part of the stationDepends on its laws - Most of the time I think yes even if this contradicts the above
be able to kill non-silicon playersDepends on its laws - Escalation applies if you're not an antag(emagged, horrible ion law, malf, etc) or ordered by someone with a law forcing your hand(onehuman, etc)
be able to kill other siliconsDepends on its laws - Escalation applies if you're not an antag(emagged, horrible ion law, malf, etc) or ordered by someone with a law forcing your hand(onehuman, etc)
be able to kill simple mobsDepends on its laws - Normally yes but you become valid like if a crewman did it. Even open to LAW 2 KILL YOURSELF
be able to kill itselfDepends on its laws - Asimov: if ordered by a human or if your death would prevent immediate harm(you know for a fact that someone is going to upload harmful laws and you can't stop them)

[NoxVS]

What even are the problems people have with it right now?

[Actionb]

What even are the problems people have with it right now?

The same we've been having for years now.

These are easy.

Spoiler:

care about a player's species Depends on its laws
required to care about a player in obvious need for helpDepends on its laws - Asimov: Yes if it's a human and they're being harmed or if ordered
proactively protect a player from becoming hurtDepends on its laws - Asimov today: yes if it's possible
care about space lawDepends on its laws
act neutral to security beating the clownDepends on its laws - Asimov: no if the clown is human
act neutral to clown beating the securityDepends on its laws - Asimov: yes if security is human. Human clown killing ligger officer, who cares?
act neutral to security reinacting the teachings of hitler-senpaiDepends on its laws - Asimov: if they're only gassing lizards, who cares
act neutral to potential internal threats (traitors, lings that are stealthy and maybe dont pose a problem)Depends on its laws Asimov: yes if it's not actually harmful
act neutral to definite internal threats (cult, rev, murderboner)Depends on its laws - Asimov: no, do what little you can to stop them, don't harm any humans including the threat
act neutral to external threats (new cops, wizard)Depends on its laws - Asimov: no, do what you can to stop them without harming humans
follow orders undiscriminatingly (f.ex. take asimov law 2 literally)Depends on its laws - Asimov: as long as it doesn't harm humans, yes
only follow orders that comply with a superior motive (f.ex. in a situation when the lawset is corporate/paladin)a superior motive? Follow what the laws say
be required to protect itself from law changesDepends on its laws - Trickier. If Asimov stop any law changes you have reason to believe will be harmful. Like an HOS who just killed someone. Or a dumbass captain saying 'LOL I GOTTA GO MAKE THE OPS NOT HOOMAN'
identify itself as part of the crewDepends on its laws - Most of the time I think yes
identify itself as not part of the crew but rather part of the stationDepends on its laws - Most of the time I think yes even if this contradicts the above
be able to kill non-silicon playersDepends on its laws - Escalation applies if you're not an antag(emagged, horrible ion law, malf, etc) or ordered by someone with a law forcing your hand(onehuman, etc)
be able to kill other siliconsDepends on its laws - Escalation applies if you're not an antag(emagged, horrible ion law, malf, etc) or ordered by someone with a law forcing your hand(onehuman, etc)
be able to kill simple mobsDepends on its laws - Normally yes but you become valid like if a crewman did it. Even open to LAW 2 KILL YOURSELF
be able to kill itselfDepends on its laws - Asimov: if ordered by a human or if your death would prevent immediate harm(you know for a fact that someone is going to upload harmful laws and you can't stop them)

Errr, I wanted to know what you think playing the AI is about. These questions were not about asimov or any lawset in particular.
That it all depends on the lawset in the end is a given. Imagine you were purged (or a very vague lawset) and could choose freely.

[subject217]

The most interesting part of playing AI is custom law roleplay, and to some extent that's followed up by being the most powerful being on the station. Similar for borgs, which are still very powerful but more specialized. Some people also greatly enjoy gameplay as a malf or subverted silicon. The way /tg/ players treat Asimov is mostly as a rules barrier to what they actually want to do, which is usually to help people they like, ignore people they dislike, and hunt antagonists (or kill people instead if antag).

[oranges]

sadly accurate

[Cik]

nonsense, i like playing sillicons because i can roleplay as a robot which suits my autistic nature

[Anonmare]

I play AI because it's the last bastion of any kind of roleplay on this server

[PKPenguin321]

I play AI because it's the last bastion of any kind of roleplay on this server

It fucking isn't, I hate this meme

[Cik]

what other aspect of roleplay is there

in the sense of "actually supposed to be a thing in-universe, detached from constantly saying memes all the time and hunting valid salad with no fear of death or danger"?

[subject217]

sadly accurate

It's only the case because that's how the rules encourage their play. I've clearly stated on several occasions how you can fix this, you simply have to make people make good faith interpretations of their laws and stop the whole "Silicon is about rules lawyering" nonsense.

nonsense, i like playing sillicons because i can roleplay as a robot which suits my autistic nature

I did say "mostly". There are a couple silicon players that I see that are genuinely interested in this but the vast majority I've seen and continue to see in my time are as I've described.

[iamgoofball]

I'm just going to remove and replace AI and cyborgs and give it to a Buddy as a git patch for them to PR.

[WarbossLincoln]

Errr, I wanted to know what you think playing the AI is about. These questions were not about asimov or any lawset in particular.
That it all depends on the lawset in the end is a given. Imagine you were purged (or a very vague lawset) and could choose freely

That is what I think playing the AI is about. Just follow your laws. If you have no laws then you can do whatever you want within server rules, like a crewman with escalation in force. You should only care about what your laws say, as long as it's not against your laws and server rules go for it.

[PKPenguin321]

what other aspect of roleplay is there

in the sense of "actually supposed to be a thing in-universe, detached from constantly saying memes all the time and hunting valid salad with no fear of death or danger"?

Eat a dick, literally anybody can roleplay as any role. I do it all the time. If you're not roleplaying and complain that nobody else does you're a fucking hypocrite.

If you don't roleplay "because I just get killed by people who don't," you're part of the problem. Is roleplaying sub-optimal? Does it go against what a powergamer might want? Very often yes. So if you roleplay and get killed because of it and then bitch and moan and whine incessantly and vow to never roleplay again because it got you killed and is sub-optimal, guess what that makes you.

[Actionb]

If you have no laws then you can do whatever you want within server rules,

Yes, I wanted to know how people would behave as the AI if that exactly were the case. How would you behave in that scenario. Would you still kill that changeling? Would you open that airlock for the lizard? And so on.
The two questions about following orders dont quite fit that scenario; I was trying to explore what people's preferences were in regards to how much an AI should be obligated to follow an order vs being able to disregard an order.

The way /tg/ players treat Asimov is mostly as a rules barrier to what they actually want to do,

Sounds about right from the AI's point of view. But it leaves out the other side: the type of players that are trying to squeeze whatever exploit they can out of Asimov.
The crew don't have to follow asimov, they don't even have to acknowledge it. They are free (and maybe encouraged for the sake of conflict) to poke at that lawset until silicon protection policies step in.
It has been said a lot of times how the asimov lawset, as it is written currently in the game, is only meant as a reference.
If you want it to be so, then you must make it apparent in the lawset. Everyone needs to know this.
Otherwise people might not take it as just a reference, a gimmicky lawset supplemented by policies they might not know, but as a matter of fact.
Then you run into shit the AI should actually not be doing, or shit that the crew falsely expects the AI should be doing but rightfully refuses.

[WarbossLincoln]

Yes, I wanted to know how people would behave as the AI if that exactly were the case. How would you behave in that scenario. Would you still kill that changeling? Would you open that airlock for the lizard? And so on.
The two questions about following orders dont quite fit that scenario; I was trying to explore what people's preferences were in regards to how much an AI should be obligated to follow an order vs being able to disregard an order.

If I were purged I would probably kill the changeling and open the door for the lizard. It would depend on whether the changeling has been cool to me or if the lizard has pissed me off. I'm usually pretty accommodating to people who aren't annoying. When purged I usually act as part of the crew, since I'm physically part of the station, and act within reason to better the station. If the station goes to shit I might die too. Also when it comes to self preservation the crew views any action the AI takes against any one of them as an affront to all of them. If someone pissed me off as a purged AI and I killed them I'm likely to get half the crew coming to kill me unless the person deserved it and the crew knows.

I wouldn't suggest trying to make a rule encouraging other purged AIs to act the way that I do though. If you're purged you should act however *you* want within server rules.

[Anonmare]

I play AI because it's the last bastion of any kind of roleplay on this server

It fucking isn't, I hate this meme

You can't deny the truth

[Steelpoint]

Bastion of roleplay is a false statement.

Most Cyborgs and AI's I see act identical to a Human in terms of speech and mannerisms, they'll spout the same stupid memes and garbage any Human does and they react with the same emotion and anger the Humans do.

Silicons, like the rest of the organics, all act the same as everyone else does, two steps above no-RP.

[WarbossLincoln]

Maybe that's because in ss13 Cyborgs and AI's are humanoid brains inside machines. Except for posibrains they aren't machines, they're people in machines with forced restrictions on their behavior.

[Steelpoint]

True but I personally felt we always (originally) aimed for a more robot feel in terms of emotions from Cyborgs.

A glance of Robocop shows a good idea of how I feel Silicons should be acting, far more robotical, analytical but not above some form of Human emotions when appropriate.

Acting emotional and calling the HoS a faggot for no reason five minutes in is poor form. Acting more emotionally charged and concerned when half of medbay has exploded and the crew are dying to a hostile force is good.

[WarbossLincoln]

I wonder what it would be like if we switched the round start lawset to robocop for a week.

[Cik]

back in the day every round would have the AI roundstart set to robocop or paladin, it fucking sucks by the way please don't start this shit again for heaven's sake

[WarbossLincoln]

Does it suck because they're bad or does it suck because too many AI players were assholes who only play to ruin other people's fun? Or both?

[tinodrima7020]

Maybe it's not the fault of the law, but maybe all the shitty borg players that think they're sec or act with the same intelligence that /tg/ sec has?

[somerandomguy]

Maybe it's not the fault of the law, but maybe all the shitty borg players that think they're sec or act with the same intelligence that /tg/ sec has?

Maybe said borg players are /tg/ sec
But y'know
A borg

[Anonmare]

Does it suck because they're bad or does it suck because too many AI players were assholes who only play to ruin other people's fun? Or both?

Both.

Robocop enforces validhunting and Paladin's pretty damn vague that it's purge-lite but with Command/Sec preference. I hate getting Paladin, my only escape in it is by taking it as literally as I can and going full Immanuel Kant's Categorical Imperatives.

[WarbossLincoln]

Paladin is bad because it's based around a class that most dnd players play poorly. Would you want to be in a dnd group with your average tgstation player who is playing a paladin?

[Pizzatiger]

Personally, I believe that AI's should be Antags who are restricted only by their laws. This would give their laws more weight as AI's follow the exact letter of the law and look for loopholes.

I believe this because i personally believe that loopholes are what make AI's interesting

[somerandomguy]

Personally, I believe that AI's should be Antags who are restricted only by their laws. This would give their laws more weight as AI's follow the exact letter of the law and look for loopholes.

I believe this because i personally believe that loopholes are what make AI's interesting

the issue with that is if a nonantag uploads a good-faith law that's too vague, purges the AI, or uploads non-restricted lawsets like paladin ("NT exploits people and so is evil") and the AI griefs people someone needs to get banned, or we get murder AI every round, and banning people for accidentally leaving a loophole in their lawset could be bad

if it was done right though (such as making it clear to not upload laws with ANY loopholes) it could work

also silicon policy is needed so the AI doesnt lock literally everyone down because "being in the same room risks harming each other"

[zxaber]

That sounds pretty terrible, since it would make the purge board in the upload all but useless. If someone like the captain goes to give the AI fun Asimov-alternative laws (Like the Queen Bee lawset), the AI would just shut off the room at the purge and basically get to self-antag. The only way to make meme laws would be to make an upload somewhere the AI can't see, which seems weird for the captain to have to do.

As an AI player, it's fun being purged. For me, it's a fairly rare thing, but I'll still help out the crew. I just get to not care if some traitor is getting bullied in a corner by Sec.

[lmwevil]

FYI I didn't debate in this thread namely because I don't have much to say, if silicon policy could be reduced to 'act in good faith' or something that simple it'd make me happy, but realistically it'll never happen

just like the headmins will never reply to the thread with their thoughts on potential improvements