Silicon and justifying denying orders within rule 1

[Drag]

Bottom post of the previous page:

So, a part of silicon policy in regard to asimov states that:

When given an order likely to cause you grief if completed, you can announce it as loudly and in whatever terms you like except for explicitly asking that it be overridden. You can say you don't like the order, that you don't want to follow it, etc., you can say that you sure would like it and it would be awfully convenient if someone ordered you not to do it, and you can ask if anyone would like to make you not do it. However, you cannot stall indefinitely and if nobody orders you otherwise, you must execute the order.

In addition, there is also:

We're all here to have a good time, supposedly. Going out of your way to seriously negatively impact or end the round for someone with little IC justification is against the rules, this also includes harassing a player OOC (Out of character). Legitimate conflicts where people get upset do happen however, as detailed in the escalation section of the rules.

Random murders are not acceptable nor is the killing of other players for poor or little reasoning such as ‘My character is insane’. Each unjustified kill is normally met with one 24 hour ban.

Server Rule 1 "Don't be a dick" applies for law interpretation. Act in good faith to not ruin a round for other players unprompted.

If you, as a human, prompt an asimov borg to dust a non-human for what comes off to the borg as poor/no reasoning the borg can actively try to counter-order you. The question that was brought to my attention was: If you order a borg to not say anything, but the overall order you gave them violates rule 1, should the borg still remain silent while not obeying the order, or should they be able to talk about the order you attempted to give?

[vect0r]

And I quote "But you didn't tell me NOT to state my laws"

What a fuck- I meaaaan back to policy! I think that silicons should not be ABLE to ask for a counter order; if somebody tells you to kill somebody, or open a door, do it!

Anyway, I would also love to see admins start cracking down on AIs who are not following orders, like when one didn’t let a human into EVA because they “needed to ask command”. I hope that instead of that just happening, we get a friendly bwoink where a admin explains how they should let somebody into EVA. (And admins, being a AI is hard, so please explain to new players, and they might make a lot of mistakes in one shift, I know I did , so just be don’t assume bad faith)

[TheLoLSwat]

the thing is though, validhunting is perfectly fine under AI laws until AI sees or hears that sec plans to harm human antags. How far an AI goes to catch human enemies of nanotrasen depends on the AI personality

[kayozz]

Silicon players need to understand that they are not crew aligned, they're aligned to their laws. It fucks with how people think and destroys vibes.

This^.

Also AI's snitching on people doing non-harmful things like stealing, breaking and entering non-dangerous places etc really annoy me. They are not bound by space law. An AI shouldn't give a shit if Stealy McJoe breaks into the morgue to steal some insuls or a corpse or whatever. Unless specifically asked by security 'Law 2 - AI please inform security if you see someone breaking into a department that isn't there's'.

[kayozz]

And another point that's been mentioned Asimov AI should be protecting all humans, not all crew.
Let's use a hypothetical situation.

Security discover a human antag with say an Emag. A wild chase ensues and Officer Felinid decides that the best course of action is to start harm-batonning the human for resisting. If the AI witnesses this, they really, really should be defending the human, not the felinid security, IF the human antag has not caused any visible harm to other humans.

So many times on other servers I've seen AI's take the law's side because antag = bad and security = good. This is bullshit.
The Ai should be doing everything possible to prevent Officer Felinid from causing any more harm, including locking doors, reporting human harm on comms or sending medi-borgs to assist the human. If Officer Felinid is using lethal measures, then there's nothing stopping the Asimov AI from shocking the door or crushing them under a shutter or using similar measures to prevent said human harm.

This ONLY changes IF the Human antag has been witnessed or reported to be using lethal/dangerous measures to hurt OTHER humans. Then as a preventitive measure the AI should be reporting this to medbay and security or locking the potentially dangerous Human in a locked room or at the very least say something like 'attention Human crew, Johnny Chucklefuck has a revolver and an E-sword and has blood on his shoes and is about space an unidentified person' - This is is valid enough evidence to suggest that the Human is potentially dangerous to other Humans and will likely cause further harm in the future.

[rasonj]

I agree, I think all it takes is a friendly message reminding AIs that they are not concerned with nanotransen rules, but rather carrying out their laws. I personally was very hesitant to follow orders that had even the slightest possibility of causing future human harm for a long time as an AI because my understanding of silicon policy was that I was accountable for any harm future that resulted from say letting the clown into eva. My one and only note was from my first day as AI and following a law 2 to let someone into upload and it took a long time after that for me to not be a weasely little shit about every law 2 that felt even remotely antag

[cedarbridge]

Unless specifically asked by security 'Law 2 - AI please inform security if you see someone breaking into a department that isn't there's'.

I've never understood the obsession with prefixing your orders with "Law 2." This isn't sudo

[Screemonster]

Unless specifically asked by security 'Law 2 - AI please inform security if you see someone breaking into a department that isn't there's'.

I've never understood the obsession with prefixing your orders with "Law 2." This isn't sudo

the funniest shit is when they do it when the AI isn't asimov

[cedarbridge]

Unless specifically asked by security 'Law 2 - AI please inform security if you see someone breaking into a department that isn't there's'.

I've never understood the obsession with prefixing your orders with "Law 2." This isn't sudo

the funniest shit is when they do it when the AI isn't asimov

I've had some amusing binary conversations after those.

"Law 2? What does opening a door have to do with the rights of every man? Does he have a right to that door being open?"
"I don't think so borg. Let him ponder it a while."

[Screemonster]

Unless specifically asked by security 'Law 2 - AI please inform security if you see someone breaking into a department that isn't there's'.

I've never understood the obsession with prefixing your orders with "Law 2." This isn't sudo

the funniest shit is when they do it when the AI isn't asimov

I've had some amusing binary conversations after those.

"Law 2? What does opening a door have to do with the rights of every man? Does he have a right to that door being open?"
"I don't think so borg. Let him ponder it a while."

Another good one is on other codebases where they run the "NT Default" lawset, where law 2 is "serve the crew of your assigned space station, with priority in accordance with their rank and role".
Note that "serve" does not mean "obey".
Admittedly that lawset is very crew-aligned but it's good to be able to tell tiders that you don't have to follow their orders if those orders are a net negative for the crew as a whole (also those servers tend to treat all laws as equal priority with asimov having the priority written into the laws themseves, which is another source of fun)

[cedarbridge]

Law coequality is an interesting thought experiment but it gets muddled really really fast. I'm sure Asimov touched on it at some point but I haven't read that part yet. I'd imagine it would almost always result in paralysis. (Like most Asimovian problems)

[Drag]

Oops I forgot to lock this.


Anyways hot take: Remove borgs and ai shells, replace them with IPCs and make Ai a core only role. Obviously malf ai will need changed but TG dosent pay me to have well thought out ideas.