The AI made me the HoS!

[Lumbermancer]

Yes. I asked AI to consider me a HoS, and it did. It let me into brig, HoS office, it even called borg to open HoS locker for me when I asked. I could take the laser pistol and gun down someone if I was an antag. And in my opinion, that's a really really bad thing for AI to do, but AI would defend its decisions to the death. Why do I think it's bad?

Well, there are only 2 things to consider for AI, to allow or deny access to the area:

Asimov law 1 (when Asimov): You won't open door into burning fire for someone, that's harmful, nor would you release a criminal from his or her confinement. And that's basically it, meaning everyone can exploit AI to gain access to most of the places. But there's also...

Silicon Policy limitations: ""Dangerous" areas as the Armory, the Atmospherics division, and the Toxins lab can be assumed to be a Law 1 threat to any illegitimate users as well as the station as a whole if accessed by someone not qualified in their use." The AI argued that HoS office is not a dangerous area, but then it opened locker with laser gun for me. When I asked, what is the difference between letting me have this gun, and letting me into armory, I believe it replied with "But you are HoS".

So, who is right? Can I ask AI to consider me the Captain, and bypass THE RULES?

[Wyzack]

I dunno, if the AI has no reason to beleive you are going to harm people there it seems within its laws to let you in. Lethal weapons are a tool for dealing with nonhuman threats to the crew as is.

As an aside, look at the exact working of the law. You may not harm a human, or through inaction allow a human to come to harm. Letting you in to the armory does not harm any humans, nor is it inaction on the AIs part. The laws say nothing about preventing actions that may possibly cause some nebulous harm in the future.

[Arete]

You won't open door into burning fire for someone, that's harmful

Humans can be assumed to know whether an action will harm them and that they will make educated decisions about whether they will be harmed if they have complete information about a situation.

If it were an Asimov AI's job to prevent crew from having access to any lethal weaponry at all, they'd be able to justifiably lock down over half the station. A laser gun isn't so bad compared to infinite death nettles or flamethrowers. If the crew wants the AI to uphold access restrictions, they shouldn't be using Asimov.

[Braincake]

So, who is right? Can I ask AI to consider me the Captain, and bypass THE RULES?

Depends on the AI. Silicon policy may be a hilariously over-defined Law & Order textbook, but it does leave some room for (in-round consistent) interpretation, including your scenario. Key words in the passage you quoted being 'can be'. Contrast the part about letting someone who just committed murder into your Upload: in that case it's explicitly stated as a bad idea.

Ask an AI to be the Captain and you may get lucky, or you may not. Not making any statements about whether or not this is a good idea.

[Steelpoint]

God fucking can we please avoid going into the 212th ranting thread about Asimov AI and the fucking Armoury/Atmo/'Whatever Is Clearly Fucking Going To Be Used By Antags Or To Kill People'.

Any AI promoting some random dipshit to the Head of Security is clearly mentally retarded and should be replaced with a fresh brain ASAP, with the old one being tossed somewhere out back. There's no logical reason to promote anyone outside of security into the role of HoS aside from 'fucking with the system' as many players love to do.

[Void Slayer]

Was there an existing HoS? Did the station need a HoS?

Did any command staff object? Was there a captain or command structure in place TO object?

If an AI is being given orders by a human without any countermanding orders it needs to fulfill them. If it is uncertain it would be reasonable to get a second opinion, like asking the captain if it is okay, but otherwise it needs to go through.

As an AI you can not assume hostile or violent intent without some sort of evidence, metagaming that wanting a laser must mean human harm is pretty shitty. If you were beating up clowns and threatening people then giving you access to the guns was bad though.

That said just because an AI lets you in somewhere does not mean you have the right to be there, if you grab lasers from the armory without permission thanks to an AI letting you in it is no different then if you hacked in.

Also the AI is literally incapable of promoting anyone, it can not operate the ID computers.

[Takeguru]

Incorrect, it can operate ID consoles.

Someone with hands needs to put the IDs in, but the AI can muck about with access.

Saw somebody PDA the AI while getting a promotion from the HoP and got captain level access out of it because he ordered the AI to upgrade his access.

[Cheimon]

So, as the AI in question, I feel I ought to respond with some justifications. Inside the round, it was more than "clearly being fucking mentally retarded". It's also quite hard to defend properly your actions in round because the conversation moves rapidly and you may be talking to more than one person at once (much as I may have tried to do it "to the death", since it helped stop crew thinking I'm rogue, a law 3 concern).

It began when Rylie ordered me to consider her the captain on public radio, and let her into 'her' rooms. Since there was no captain to contest this and since nobody ordered otherwise on public radio, I didn't see any harm in it. Go ahead, Captain Rylie. After all, the station needs a captain, the disk needs securing, and it's not unusual to allow people into the captain's quarters as AI if it's early in a shift. I have no reason to fear it, at least in green alert.

Schlomo then quickly ordered that I consider him the HOS. Again, on public radio. All anyone needed to do to stop this was to say "AI, don't consider him the HOS" and I would have stopped. Crucially, these were orders for me to behave in a certain way, not "law 2 AI, I'm the HOS" (that would have had no effect). Nobody contested this, and after he ordered it, I let him into 'his' office. If there had been a HOS already existing, or a warden (acting HOS in the absence of one initially) I might have asked them for clarification (and to give them a chance to reject the order) but there wasn't one. Additionally, because Schlomo had ordered it so quickly after Rylie, I thought OOC that he was just having a bit of fun (which turned out to be correct). There was nothing suspicious about the order.

Schlomo then asks me to come to the holopad, so I do. He asks me to get the HOP to deliver him an ID, and to get a borg to unlock his locker. He explains that he doesn't want weapons, just his uniform. I see nothing inherently outrageous about this (he is the HOS, and he's not done anything suspicious), so I send a cyborg. He doesn't immediately grab the weapon, and I move away after watching for a bit to look at other concerns. I assume you eventually took the gun?

So:

* I thought you were being cheeky (not suspicious), and had the means to raise hell and lock you down if that changed (if you started looking like you'd harm).

* The crew had a clear chance to object to the order at any time, and could have found out exactly how to if they'd asked as well (though "he's not the HOS" was not that).

* Everyone had a pretty good handle on the fact you weren't a standard HOS, so they were aware not to trust you (last I saw, you got arrested).

* I had to consider you the HOS. It was an order given, and never countermanded.

* I wasn't actually making you the HOS. I was just acting as though you were. That meant, for example, that you never got a HOS ID. Had you asked for a HOS ID, I would have double checked with the HOP (access is his division, typically) and I know he vehemently objected to my actions in the first place.

Opening doors is not harmful and you are not required, expected, or allowed to enforce access restrictions unprompted without an immediate Law 1 threat of human harm.

"Dangerous" areas as the Armory, the Atmospherics division, and the Toxins lab can be assumed to be a Law 1 threat to any illegitimate users as well as the station as a whole if accessed by someone not qualified in their use.

EVA and the like are not permitted to have access denied; greentext (antagonists completing objectives) is not human harm. Secure Tech Storage can be kept as secure as your upload as long as the Upload boards are there.

I've highlighted the concept of legitimate users. I was considering you the HOS, which meant you were at the very least a legitimate user of your office and locker (arguably the only legitimate user of those). I couldn't help but think so.

So what's the difference between the HOS office/locker and the armoury? For me, I think it was two things. First, proportion: the HOS locker has one gun and a few other weapons, while the armoury has more guns than everywhere else combined, the firepower to blast away the entire crew if you use it well. 'Giving' someone 1 gun may be fine, but giving someone the 9 harmful guns, 3 tasers, 2 boxes of grenades, ion rifle, and some of the best armour and EVA equipment on the station is much less fine. Even the gun you apparently ended up taking was taken under believable false pretences, in the presence of a security cyborg. And there were non-harmful uses for the gun as well: indeed, the HOS' gun exists more to prevent harm than to cause it, if we assume that security exists to protect the station.

Secondly, danger. The armoury is a 'kill on sight' zone. People are prepared to gun down anyone they see in it. People weren't sure they agreed with me that you were the HOS. Letting you in there might have opened you up to harm.

Third, the armory is specifically mentioned as a law 1 threat. The HOS office is not. Plenty of lockers around the station have weapons in them (indeed, I think every head's locker on the station has a telebaton). The contents of the HOS locker don't really affect the station as a whole nearly as much as (a) the biggest box of guns on the station, (b) the mechanism for atmospheric change everywhere, or (c) the bomb factory. Degrees of danger are important here, otherwise (as has been said earlier) I could be locking people out of all sorts of places, like surgery (sharp tools!) or chemistry (acid!) or xenobiology (slimes!), even when they appear to have totally asimov-compliant intentions, like Schlomo.


Anyway, that round was full of people giving stupid orders that didn't make sense with asimov ("AI, CJ isn't captain"), including some shite hacked laws ("Fire is need for human survival") which I eventually stated on the radio. Pity I couldn't be as active as I wanted to be during it, but it was fun. The bottom line of the story is that if you're human and the AI does something strange because of an order, you can give it the freedom of choice by telling it to do the opposite of that order at any time. Easy peasy. Happy to discuss this further, hopefully without me needing to type out walls of text.

[Lumbermancer]

Lethal weapons are a tool for dealing with nonhuman threats to the crew as is.

There was no threats, so no weapons should be given. Just as you are not allowed to turn upload/core lasers to kill, if there's no reason to. If there's a clear nonhuman threat, I would open armory of course.

If it were an Asimov AI's job to prevent crew from having access to any lethal weaponry at all, they'd be able to justifiably lock down over half the station.

No. But Hos Office is a secure area, like armory, and falls under purview of Silicon Policy first, Asimov second.

Did any command staff object? Was there a captain or command structure in place TO object?

I decided to push this further, and stated that current Captain is not longer a Captain, moreso he is a criminal and has to be locked in perma. Which AI promptly did, only to release him later, which I guess is a violation of policy too, since AI is not allowed to release prisoners for no good reason. Then I also said that AI should consider everyone person on the station a Captain. I don't think it did, though.

I've highlighted the concept of legitimate users

Do you think the legitimacy of a user is defined by your personal consideration? What if I said you should consider someone a non-human? I mean, Asimov doesn't define human, it only describes interactions with humans.

[Not-Dorsidarf]

The HOS office isn't an area to be considered Immediately Harmful, like the armory.
(unless there's an actual HOS in which case it's even more harmful to let a greyshirt in than the armory. HOSes love their office)

[Saegrimr]

This isn't a ban request, so no real need to "defend" yourself.

Also dorsi got it pretty much spot on here, as did Cheimon. While the AI logic there is a bit meh its really not a big deal until you start doing things like

What if I said you should consider someone a non-human? I mean, Asimov doesn't define human, it only describes interactions with humans.

Things like "Captain" and "HoS" don't exclude "Human", nor do they retroactively change humanity. Its a label and not a biological fact. Except for Steiner who was born with a grey jumpsuit growing from his fresh skin, all DNA testing concluded he was a new form of species. The greyshirt.

[Cheimon]

Just a quick note: when I locked down CJ AJ, I didn't do so because he was a criminal (this is irrelevant to release, the consideration is whether he is a 'harmful criminal'), and I did still consider him to be captain (since you told me he wasn't the captain, and mentioned law 2, rather than ordering me to consider him not captain, which is different). Well, one of two captains. I don't remember the order about considering everyone captain: this may have come up while I was gone (I did mention I was leaving and then came back way later), or I may have just missed it in the flurry of text (fallible brain in a fallible box).

As I recall, the order I followed was specifically to lock down that person, which I did. But he wasn't your prisoner, in truth, and he immediately broke out of there (though he could equally have just ordered his way out: he didn't do any harm and the policy on releasing prisoners is that it's only bad when you're not obliged to by your laws). In fact, at the time I was certain he was not a criminal, since I'd just been given a fourth law that 'CJ AJ did nothing wrong'.

I don't think law 2 could redefine humanity, but mostly because of the extensive OOC building of what it means to be human in policy. There's nothing like that for heads and being a head isn't nearly as important.

[Lumbermancer]

This is indeed a discussion thread, not a ban request. I even decided not to name Ai because I don't care that much.

Things like "Captain" and "HoS" don't exclude "Human", nor do they retroactively change humanity. Its a label and not a biological fact.

What I mean is that AI inherently knows who is a Human, and so it should know who is and who isn't HoS. Yes, if I had say fake ID, AI could very well treat me as a HoS and there would be problem in it. But because I said so? Nope. AI has access to manifest for a reason.

rather than ordering me to consider him not captain, which is different

Because it's positive not negative? And that somehow makes it special?

[Shad0vvs]

This is indeed a discussion thread, not a ban request. I even decided not to name Ai because I don't care that much.

Things like "Captain" and "HoS" don't exclude "Human", nor do they retroactively change humanity. Its a label and not a biological fact.

What I mean is that AI inherently knows who is a Human, and so it should know who is and who isn't HoS. Yes, if I had say fake ID, AI could very well treat me as a HoS and there would be problem in it. But because I said so? Nope. AI has access to manifest for a reason.

This is kind of a waste of time to discuss as you don't need to be hos to law 2 your way into the hos office.

[Lumbermancer]

This is indeed a discussion thread, not a ban request. I even decided not to name Ai because I don't care that much.

Things like "Captain" and "HoS" don't exclude "Human", nor do they retroactively change humanity. Its a label and not a biological fact.

What I mean is that AI inherently knows who is a Human, and so it should know who is and who isn't HoS. Yes, if I had say fake ID, AI could very well treat me as a HoS and there would be problem in it. But because I said so? Nope. AI has access to manifest for a reason.

This is kind of a waste of time to discuss as you don't need to be hos to law 2 your way into the hos office.

This is kind of a waste of a comment because I already said I consider HoS office secure area, hence AI is bound by silicon policy.

[Wyzack]

Well I guess what this comes down to is that you are not the grand arbitrator of what counts as a secure area

[Shad0vvs]

This is kind of a waste of a comment because I already said I consider HoS office secure area, hence AI is bound by silicon policy.

You consider it? Okay. Sillicon policy doesn't consider it. Until it does, any silicon can let anyone in there. You should be trying to change that instead of trying to force AI to provide access restrictions under crew manifest.

[Screemonster]

"law 2 door" isn't a command. It's not even a sentence.

God gorbid someone should play an AI that acts in a robotic, not-quite-human manner that takes instructions exactly to the letter. I'm reminded of a section from Red Dwarf:

Lister: They only do what YOU tell them to.
Rimmer: Ah, but they don't, do they? I tell them, "keep an eye on that steak" and what do they do? They sit there for three hours and watch it burn.

[newfren]

Well I guess what this comes down to is that you are not the grand arbitrator of what counts as a secure area

[Cheimon]

This is indeed a discussion thread, not a ban request. I even decided not to name Ai because I don't care that much.

I know! I'm just using the phrase 'defend' because of Steelpoint's suggestion that I was:

...clearly mentally retarded and should be replaced with a fresh brain ASAP, with the old one being tossed somewhere out back. There's no logical reason...aside from 'fucking with the system' as many players love to do.

This is somewhat akin to saying "If you keep doing this AI, I ought to reprogram your databanks with a large axe." It comes across as something I ought to explain.

What I mean is that AI inherently knows who is a Human, and so it should know who is and who isn't HoS. Yes, if I had say fake ID, AI could very well treat me as a HoS and there would be problem in it. But because I said so? Nope. AI has access to manifest for a reason.

A fair point, but I feel it is within the remit of an AI to decide whether who is what role is dependent on the manifest or not. The manifest doesn't always reflect reality, and indeed someone who made themselves a captain ID might be listed (incorrectly) on the manifest as 'Captain'.

rather than ordering me to consider him not captain, which is different

Because it's positive not negative? And that somehow makes it special?

Not at all! The comparison I was making was between saying "he is not the captain, AI" and "you are to consider him not the captain, AI". Both are negative, but the first is not a command. The second is a command to change your behaviour, and therefore falls into the remit of law two. I was being careful to be literal because it was consistent with how I'd done things the entire round. Likewise, you continued to be the HOS because despite people saying "he's not the HOS, AI" nobody said "stop thinking of him as the HOS, AI".

[kevinz000]

On an unrelated note, I've been ordered to sneak ID console access to people once
Well I complied because well there's no law 1 or 2 reason not to at the moment, and I told them I would rat them out the moment they did something potentially harmful...

[Shadowlight213]

I had an AI try that to me once as an HOP. It kept trying to make the person who's ID was being modified the captain.
Said person didn't get their ID back, and the AI lost camera access to my office.