Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jhxmt]

The logic of the Three Laws in Asimov's books isn't particularly easy to define outright, as most of his whole Robot sequence was intended as an exploration of what those seemingly-simple three laws could actually allow/imply - which would seem to indicate that there's isn't necessarily a 'correct' interpretation of how the laws would cause an AI to act.

On the whole 'doing a small bit of harm to prevent a larger bit of harm' logic, it's definitely not clear cut:

Spoiler: Spoiler from I, Robot's 'The Evitable Conflict' (click to show/hide)

From the wiki page on The Evitable Conflict (as I don't have my copy handy):
[The protagonists] discover that the Machines have generalized the First Law to mean "No machine may harm humanity; or, through inaction, allow humanity to come to harm." (This is similar to the Zeroth Law which Asimov developed in later novels.) Dr. Calvin concludes that the "glitches" are deliberate acts by the Machines, allowing a small amount of harm to come to selected individuals in order to prevent a large amount of harm coming to humanity as a whole.
  - https://en.wikipedia.org/wiki/The_Evitable_Conflict

This would seem to imply that, at least in Asimov's interpretation of them, the Three Laws can allow for AIs to permit, even to cause, 'lesser' harm to prevent 'greater' harm.

But, as others have implied, gameplay and enjoyment might be greater considerations anyway. 

[deoloth]

But the AI(Assuming it is Asimov) isn't allowed to cause any harm at all, even if causing harm can prevent greater harm.

Honestly, I think it needs a bit of flexibility with that. The AI is weak enough as it is.

Well, its not like hostage situations are particulary annoying for the AI anyway. It can afterall just turn off the radios in its chamber, therefore not knowing what the hostage taker is saying, which means the AI doesn't know if he is going to cause harm to the hostage if the AI doesn't do so and so.

This solution bumps up against the "or through inaction" clause of the law, though.

Right here. Nail on the head.

Letting someone go who might, hell gives you good reason to believe he will, cause more harm is inaction. At that point its attempting to minimize the amount of harm. If its trying to stop all harm possible then the A.I. would just never let anyone out of the damn dorms. Shocking a hostage-taker seems like a reasonable route to take if you believe he has a good chance of inflicting greater harm on other humans.

As for orders. I believe in a Issac short story (If anyone knows which one, I would appreciate the title. I want to say 'Evidence', but I am not sure if that is right.) It was established that figures of authority hold a higher priority for the follow through of orders.

Ill admit, I only play on TG but generally I find playing Asimov as literally as possible to be the most fun, I will follow every order given that doesnt conflict. As for shocking the door, I personally think that should be a silicon jobban.

For me my biggest peeve with AI's are those that bolt things like EVA because Hurr Durr Spacesuits are harmful to humans Burr.

I would have just followed the hostage takers orders and kept security in the loop.

We had a game a few months back where an A.I. (non-traitor or malf) decided to bolt every door to the armory and security office because they would cause harm while a traitor was running around after killing someone(s?). One of the admins on at the time gave him a warning about it in OOC. So typically an obnoxious A.I. player will be warned or at least checked on this server. I think it is likely that the A.I. you dealt with had a malfunctioning law that caused him to at as if spacesuits were harmful to humans and had to be barred. I actually malfunctioned once and got the Law "Computing is harmful to humans". That was so very much fun.

[Ivan Issaccs]

No its just one of those things that used to be a regular occurence on TG and still somewhat common for AI's to bolt down EVA, Tech Storage and the teleporter.
I find its much more fun for an AI to follow every request from every crew member where its not an obvious breach.

[scrdest]

IMO, the policy should be that AI has free rein in interpreting the laws as either 'ABSOLUTELY NEVER LET ANY HUMAN HARM' by-the-book AI or 'Minimize human harm in any way possible' AI. It gives them some flexibility instead of forcing them into One True Arpee To Rule Them All.

[Flying Dice]

If I hadn't been personally involved (I was the HoS/hostage taker), I would have backed BD's decision. In hindsight, I don't think it was entirely unreasonable.

What is frustrating, though, is going out of my way as a traitor to cause chaos and take a risk like holding someone hostage (especially in an AI round, which cuts my chances of escape), and then be caught because the AI went for the "through inaction" subclause despite the fact that to its knowledge I had not killed or seriously harmed anyone, after I had repeatedly stated that my intent was to leave free and alive without killing anyone, and after I had gone out of my way several times to not follow through on ultimatums (because BD was RPing a conflict breakdown and reboot rather than just being a dick and ignoring the laws).

Frankly, it's at least partially my fault for not putting on insulated gloves before checking the door. I had them, knew that it was a possibility, and knew that the AI had told the other crew to back away from the doors. But I decided to be trusting, and it bit me on the ass. Funny thing was, I pulled the Sith traitor box, so I could've gone balls-out murderboner, but I decided to do something less boring and self-indulgent. :/

It was justified, from BD's perspective. Sort of a dick move from mine, but not the sort of thing I would have ahelped over. It'd be rather hypocritical of me to get on him for using the "through inaction" clause when I've always been a big proponent of AIs using the flexibility in Law 1 to minimize harm, rather than blindly NO HARM NO HARM-ing based on the situation of the moment. I tried to manipulate the AI, wasn't cautious enough, and paid the price for it.

What I /do/ have an issue with was when you recalled the shuttle the first time, despite me being very clear that I would kill him if you did. With a lot of people, that would have been a dead hostage right there, and a clear violation of the laws.

[LordSlowpoke]

No its just one of those things that used to be a regular occurence on TG and still somewhat common for AI's to bolt down EVA, Tech Storage and the teleporter.
I find its much more fun for an AI to follow every request from every crew member where its not an obvious breach.

ais who interpret "open door" as "open door and leave it open because i didn't tell you to close it" for nanotrasen ceo

[Nienhaus]

So how many of you guys are going to be online for All Hallows' Evening?

[Nienhaus]

So how many of you guys are going to be online for All Hallows' Evening?

Maybe me.
Anyways, anyone wanna join now?

I'm on kinda.

[Ivan Issaccs]

If I hadn't been personally involved (I was the HoS/hostage taker), I would have backed BD's decision. In hindsight, I don't think it was entirely unreasonable.

What is frustrating, though, is going out of my way as a traitor to cause chaos and take a risk like holding someone hostage (especially in an AI round, which cuts my chances of escape), and then be caught because the AI went for the "through inaction" subclause despite the fact that to its knowledge I had not killed or seriously harmed anyone, after I had repeatedly stated that my intent was to leave free and alive without killing anyone, and after I had gone out of my way several times to not follow through on ultimatums (because BD was RPing a conflict breakdown and reboot rather than just being a dick and ignoring the laws).

Frankly, it's at least partially my fault for not putting on insulated gloves before checking the door. I had them, knew that it was a possibility, and knew that the AI had told the other crew to back away from the doors. But I decided to be trusting, and it bit me on the ass. Funny thing was, I pulled the Sith traitor box, so I could've gone balls-out murderboner, but I decided to do something less boring and self-indulgent. :/

It was justified, from BD's perspective. Sort of a dick move from mine, but not the sort of thing I would have ahelped over. It'd be rather hypocritical of me to get on him for using the "through inaction" clause when I've always been a big proponent of AIs using the flexibility in Law 1 to minimize harm, rather than blindly NO HARM NO HARM-ing based on the situation of the moment. I tried to manipulate the AI, wasn't cautious enough, and paid the price for it.

What I /do/ have an issue with was when you recalled the shuttle the first time, despite me being very clear that I would kill him if you did. With a lot of people, that would have been a dead hostage right there, and a clear violation of the laws.

On TG code, AI's cannot recall the shuttle.

[Cheedows]

All Hallows? Is this some event people refuse to tell me? 

[scrdest]

If I hadn't been personally involved (I was the HoS/hostage taker), I would have backed BD's decision. In hindsight, I don't think it was entirely unreasonable.

What is frustrating, though, is going out of my way as a traitor to cause chaos and take a risk like holding someone hostage (especially in an AI round, which cuts my chances of escape), and then be caught because the AI went for the "through inaction" subclause despite the fact that to its knowledge I had not killed or seriously harmed anyone, after I had repeatedly stated that my intent was to leave free and alive without killing anyone, and after I had gone out of my way several times to not follow through on ultimatums (because BD was RPing a conflict breakdown and reboot rather than just being a dick and ignoring the laws).

Frankly, it's at least partially my fault for not putting on insulated gloves before checking the door. I had them, knew that it was a possibility, and knew that the AI had told the other crew to back away from the doors. But I decided to be trusting, and it bit me on the ass. Funny thing was, I pulled the Sith traitor box, so I could've gone balls-out murderboner, but I decided to do something less boring and self-indulgent. :/

It was justified, from BD's perspective. Sort of a dick move from mine, but not the sort of thing I would have ahelped over. It'd be rather hypocritical of me to get on him for using the "through inaction" clause when I've always been a big proponent of AIs using the flexibility in Law 1 to minimize harm, rather than blindly NO HARM NO HARM-ing based on the situation of the moment. I tried to manipulate the AI, wasn't cautious enough, and paid the price for it.

What I /do/ have an issue with was when you recalled the shuttle the first time, despite me being very clear that I would kill him if you did. With a lot of people, that would have been a dead hostage right there, and a clear violation of the laws.

On TG code, AI's cannot recall the shuttle.

Oh right, I finally realized where do I recognize you from. TG. Duh. Well, that and your nick being a name of a character in Priest.

[My Name is Immaterial]

All Hallows? Is this some event people refuse to tell me? 

All Hallows Eve or, as it is commonly known, Halloween is the time when admins have to come up with something spooooooky, lest they be mocked by their player base.

[Nienhaus]

All Hallows? Is this some event people refuse to tell me? 

All Hallows Eve or, as it is commonly known, Halloween is the time when admins have to come up with something spooooooky, lest they be mocked by their player base.

My family calls it All Hallows Eve so I just gave in. Also I did something.
https://github.com/tgstation/-tg-station/pull/1664

[miauw62]

I think I may not have stated my opinion very clearly.
Hostage situations are cool.
If an AI is always going to go "hurr durr no2 and shocked doors" if somebody tries to do a hostage situation, there will be no hostage situations and more parapen+c4 and the likes.

E:
Need moar peeps on da server.

Also ass-copying got merged, so you know what to do next time the server updates.

[Ivan Issaccs]

If I hadn't been personally involved (I was the HoS/hostage taker), I would have backed BD's decision. In hindsight, I don't think it was entirely unreasonable.

What is frustrating, though, is going out of my way as a traitor to cause chaos and take a risk like holding someone hostage (especially in an AI round, which cuts my chances of escape), and then be caught because the AI went for the "through inaction" subclause despite the fact that to its knowledge I had not killed or seriously harmed anyone, after I had repeatedly stated that my intent was to leave free and alive without killing anyone, and after I had gone out of my way several times to not follow through on ultimatums (because BD was RPing a conflict breakdown and reboot rather than just being a dick and ignoring the laws).

Frankly, it's at least partially my fault for not putting on insulated gloves before checking the door. I had them, knew that it was a possibility, and knew that the AI had told the other crew to back away from the doors. But I decided to be trusting, and it bit me on the ass. Funny thing was, I pulled the Sith traitor box, so I could've gone balls-out murderboner, but I decided to do something less boring and self-indulgent. :/

It was justified, from BD's perspective. Sort of a dick move from mine, but not the sort of thing I would have ahelped over. It'd be rather hypocritical of me to get on him for using the "through inaction" clause when I've always been a big proponent of AIs using the flexibility in Law 1 to minimize harm, rather than blindly NO HARM NO HARM-ing based on the situation of the moment. I tried to manipulate the AI, wasn't cautious enough, and paid the price for it.

What I /do/ have an issue with was when you recalled the shuttle the first time, despite me being very clear that I would kill him if you did. With a lot of people, that would have been a dead hostage right there, and a clear violation of the laws.

On TG code, AI's cannot recall the shuttle.

Oh right, I finally realized where do I recognize you from. TG. Duh. Well, that and your nick being a name of a character in Priest.

Trevor Donklestoop reporting, also your the first person to ever comment on that reference in about eight years of using it.