/tg/Station 13

Should an AI be allowed to refuse or ignore law 2 orders from Comms Agent, or anyone that cannot confirm they are human through cameras?
According to most policy threads I've read, if there is any confusion over whether or not someone is human, the AI should assume they are human, however is just a human-like voice enough to force an AI to assume humanity?

Is it considered bad faith? Metagaming to assume comms agent is human? Just one more silicon policy thread for the pile.

FMPOV--I don't mind as long as they're consistent, since they can't prove if they are or aren't human.

On one hand I think it'd make sense to assume they are human
On the other comms agents just using the AI as a vehicle to inflict minor grief is obnoxious sooo I'd say it's okay.

SkeletalElite wrote: ↑Tue Nov 30, 2021 2:58 am On the other comms agents just using the AI as a vehicle to inflict minor grief is obnoxious sooo I'd say it's okay.

If the comms agent is using the AI for minor grief the AI* can fight back the same way if anyone uses the AI for minor grief, which is to announce what orders they're giving the AI*, or use law 1 to ignore the order when it's relevant.
It can also work the other way around. If the comms agent has a cool gimmick, and just needs the AI to do something minor like... turning off some lights, it can be very frustrating the AI just responds with "sorry, can't prove you're human visibly, so no"

Mothblocks wrote: ↑Tue Nov 30, 2021 2:45 am FMPOV--I don't mind as long as they're consistent, since they can't prove if they are or aren't human.

It's quite hard to definitely prove anyone is human. Even visually someone can look human despite not being human, for example lings and hardsuits. In both of those cases AIs are expected to assume humanity, aren't they? Why is it different when it comes to verbal evidence, which doesn't prove they're human, but suggests they are?

Mothblocks wrote: ↑Tue Nov 30, 2021 2:45 am FMPOV--I don't mind as long as they're consistent, since they can't prove if they are or aren't human.

I ask to verify that they are human to follow through the order since it's just a voice over the radio. It sounds human, but I would like more before I begin opening doors and doing random things for them on the station. If I have more information, like someone visiting them and also telling me they are human, I might conclude that they are human.

Gameplay wise though I feel comms agents should do a bit more to get the AI to do things for them. Like impersonate actual humans on the station and give orders, or have an agent forge medical records for them.

mstachife wrote: ↑Tue Nov 30, 2021 3:28 amGameplay wise though I feel comms agents should do a bit more to get the AI to do things for them. Like impersonate actual humans on the station and give orders, or have an agent forge medical records for them.

This. As long as the comms agent isn’t doing destructive-yet-uninteresting grief (i.e. “hey AI, destroy the ORM lol”), comms agents should be encouraged to sabotage the station and screw with its crew. The crew can respond by hunting down the comms agent, uploading new laws to the AI, or changing the AI’s lawset altogether. Emergent gameplay at its finest.

From an admin POV: I'd be fine if an AI ignored comms agent orders until they verified humanity, I'd also be fine if they obeyed them.
From a player POV: I'll do whatever option seems to make the most fun. If the comms agent is just doing boring grief I'll ignore it, if they've got fun ideas I'll go along with it

The AI has a law to obey humans, not a law to disobey nonhumans. There is absolutely no in-law justification for "Prove you're human or I won't obey you."

Back before all the species got telltale speechverbs, the default law admin ruling was along the same lines as changelings - "Assume an unknown species person is human because if you harm them/disobey them you're more likely to be breaking laws than if you don't. You need actual proof of nonhumanity to assume they arent human"

If the comms agent is telling you to do things that are in conflict with the rest of the crew's orders, deprioritise him because he clearly doesnt have the other human's best interests in mind just like you would any other human who was telling you to.
If the comms agent is telling you to do harmful things ignore him like you would anyone else.
If the comms agent is demanding you do menial bull shit adminhelp and say you're gonna ignore his menial busywork orders like you would for anyone else.

While the Comms Agent doesn't have the advantage that the crew does of "their species is listed in medical records," until and unless you have a particular reason to believe they're nonhuman (lizardly hissing on the "s" sounds, or cat memespeak "Nya-llo everyone! How are mew today!"), you really should be treating them as you would any non-crew human. This can include loudly asking "Are you SURE I should vent the escape shuttle dock? That seems like it might cause problems later!" or whatever.

But yeah, you shouldn't be asking them to prove their humanity without a reason to doubt their humanity.

Unless the AI can prove without a doubt that the comms agent is not human under Asimov, the AI must follow the comms agent's orders as long as they do not contradict their other laws.

Headmin Votes:
Dragomagol: Agree
NamelessFairy: Agree
RaveRadbury: Agree