Thread Rating:
  • 9 Vote(s) - 3.67 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[RULE CHANGE] Make overrides and precedence no longer apply to AI laws.
#16
(06-19-2022, 09:45 AM)Maegor Wrote:
(06-19-2022, 08:50 AM)nefarious6th Wrote: not for this...

You could order the AI not to shoot you, due to law 2, provided you are a head of staff. But having a Law 0 slot available by default that you can add a law to also works.

I'm saying if you had multiple laws to add because they have to be added in sequence w/o the ability to override

the people who have energy guns and barriers are heads of staff or security. if those are also the only people the AI might tenably listen to when they say not to shoot them, then it's not really opening up counterplay ability outside the subset of players with barriers and eguns
Reply
#17
(06-19-2022, 10:01 AM)LeahTheTech Wrote:
(06-19-2022, 09:36 AM)KikiMofo Wrote: Just because a law is above the other laws doesnt mean its more important than the other laws. You would still need to state in the new law that that law takes precedence.
From the wiki page on AI laws:

Quote:IMPORTANT: Do note that if there's a conflict between two any two laws and either law doesn't explicitly (e.g. in writing) override or take precedence over the other law, then you are to prioritize the lower numbered law.
Lower numbered laws always take priority, the asimov laws just state it explicitly because that's how they were in I, Robot

Now might be a good time to remove the explicit conflict resolution clauses on the asimov laws, since they aren't required and make people think that without them law 2 could override law 1

Just because a law that says, "Kill all humans" gets put into law slot 0 doesnt mean the AI and borgs can ignore the other three laws if the kill all humans law doesnt say to ignore those laws.
Reply
#18
Yeah I'd personally like to see laws take priority solely on law order, not wording. It makes more sense from a new user's perspective that law 1 is more important than law 2. It would also lead to a lot less mess ups by newer command members who put an overrides all laws law without realizing they just made the AI rogue.

Right now if you uploaded "Tacos are now humans, this overrides all laws" you'd be making the AI rogue which is not the point of the law. The point is that tacos are also humans. It just adds a lot of unnecessary knowledge that you have to know (by reading a semi-obscure wiki page) before even touching an AI module.


Also as a side idea, a bit off scope to this proposed change but I'm putting it here because I can, it would be cool to change the way law updates work in general. Make it so you have to push a button on the law rack to send the updated lawset to the ai/borgs. So if you put in a new law in slot 4, it wouldn't send the new laws to the AI until you hit the "update" button. Same with taking a law out, you could safely take out law 1 without worrying about being lethal lasered the instant you do as long as you don't hit the "update" button until after you're done sorting the laws.
Reply
#19
I agree having a law update button would be very good instead of AI and borgs instantly going rogue when you remove a law.
Reply
#20
My main issue with a law update button is that itd be really easy to input and upload a rogue law, remove the bad law from the lawrack but dont upload it, and now you have rogue silicons that appear to have normal laws, which seems really unintuitive and would proabably lead to a lot of ahelps. Id honestly just like to see the speed of welding/screwdrivering increased but remove the time to insert laws.
Reply
#21
Just make so the law get updated on a timer, so you have like X seconds to fix the laws before the AI gets to lethal you, better not have any distractions!
Reply
#22
(06-19-2022, 11:24 AM)Sord213 Wrote: Yeah I'd personally like to see laws take priority solely on law order, not wording. It makes more sense from a new user's perspective that law 1 is more important than law 2. It would also lead to a lot less mess ups by newer command members who put an overrides all laws law without realizing they just made the AI rogue.

See, I think the opposite.  I think law importance being tied to law number is completely unintuitive to a new player.
Reply
#23
(06-19-2022, 12:04 PM)Mouse Wrote:
(06-19-2022, 11:24 AM)Sord213 Wrote: Yeah I'd personally like to see laws take priority solely on law order, not wording. It makes more sense from a new user's perspective that law 1 is more important than law 2. It would also lead to a lot less mess ups by newer command members who put an overrides all laws law without realizing they just made the AI rogue.

See, I think the opposite.  I think law importance being tied to law number is completely unintuitive to a new player.

I literally can't imagine a simpler system than "If there is a conflict, the first law takes priority".
Reply
#24
I haven't played AI at all and only a little bit of silicon now and then. But even from that I've seen some byzantine laws nonsense, such as murder all monkeys, but also monkey are humans and you can't hurt humans, with all sorts of conflicting clauses nonsense.

So yeah I think I'd like simpler, clearer laws rules. No more contradicting overrides and supercedes. The issues with manipulating the actual law racks can be adressed later down the line.
Reply
#25
(06-19-2022, 01:11 PM)amylizzle Wrote: I literally can't imagine a simpler system than "If there is a conflict, the first law takes priority".

Which is already the system currently in use and it's the guiding principle for resolving conflicting laws. (It actually takes some effort to fuck up a lawset so bad that it's impossible to interpret, I've only had it happen to me once.) The fact that the current Asimov laws even bother mentioning it is I think more to do with how they were worded in the source material than necessity. Would be confusing to keep those clauses if this went through though.

This change isn't proposing a new system so much as it is preventing people from specifying exceptions to that rule, and worded like that I think it's easier to see why I don't see it as an improvement. It robs the system of any depth it can have in favour of a mechanical implementation that frankly involves a bunch of busywork.

I strongly feel that the potential for complication regarding AI laws is the core of what makes AI laws interesting. If the lawset is hard to interpret and making borgs slow to respond that's practically intended and if you don't want that, write laws better. :P
I've personally never had 5 laws at once containing overrides (next to impossible with the upload comp, with lawracks crew will start pulling modules out before doing complex overriding) Speaking from experience you tend to get given a few minutes of grace to figure out how you're choosing to interpret a lawset when it gets changed anyway. For that matter, the captain law example is kinda odd because the possibility for human error is part of what makes the dynamic work.


However, I think that if you absolutely wanted to go ahead with this then you absolutely can't have law changes take effect immediately. You can't have it take 30 fucking seconds to weld/screw/remove/place/screw/weld a module and also give the AI an alert whenever you change a module. That's an utterly clown-ass long time in which the AI can readily laser your ass or sic sec on you (and you'd be surprised how little time it takes to beeline somewhere). Make it so you open the rack to mess with it and once you close it, it forwards the changed lawset en bloc to the silicons or something.

By the way I could well be talking out my ass on this one but I think the "letting players know" part is not really going to go away. It's not uncommon for people to return after months if not years of absence, and I doubt many of them will be getting up to date with everything before they try uploading overrides.

---

On a sidenote I've seen people mention that silicons don't have to follow speech laws anymore but I can't find confirmation of that on the wiki. Seems like a bad time if that's true but not written down anywhere.
Reply
#26
silicons dont get exempted from speech laws but you're allowed to ahelp them and the general culture people have towards it is that most people dont care if you dont follow it.

the 30 seconds of busywork bat pointed out, plus just the busywork potential in having to slide modules around in general and the essentialization of removing law 1 is kind of my problem with this.

ais are going to be very hard to rogue. ais are going to be very hard to fix. the game moves too fast for a 20 minute gear-up, hack, turret process or whatever.

fixing is my bigger concern, but for roguing too:
the manudrive is in the CE locker which requires head access; device analyzer in mech lab
barrier is in sec lockers
killswitch requires a head id

so in these scenarios (and I've already run into them without this implemented and done some evil things in order to help fix laws like break a sec locker for a barrier) if you are committed to fixing the AI but not a head, you have to do some crime beyond the door hacking and b&e into the upload. I dont see this as a feature. I see it as inherently confusing during a crisis

But if we were to cut out those scenarios where any staffie could help and assume fixing the AI in a crime-free nature aside from door hacking: then with the pathways available to unrogue (relink, killswitch, or law fix), relinking requires the manudrive. Which requires heads of staff. Killswitch requires a command-level ID. Which requires heads of staff. Barrier blocking or an egun to take out a turret so you can sit in upload for a minute and shuffle law boards around takes a head of staff for the gun or sec for the barriers.
limiting the override and precedence limits the usefulness of the freeform, which limits the ability of other players to participate in interacting with the ai, either to rogue or unrogue or upload a birthday law. a rogue AI can have open season on anyone, but not anyone can have open season on a rogue AI. I dont think that's particularly interesting or good.
Reply
#27
My biggest argument is how many rounds are Asimov laws still the norm? If you answer anything other than the majority I would be surprised.
Reply
#28
IMO roguing/resetting laws should always be easier & more convenient than kill-switching or beating the AI to death (outside of severe FUBAR edge cases) and rules + mechanics should support that

rule change is intuitive but further consideration could be given to preserving the cat and mouse game of law changes and mechanically discouraging silicon round removal



getting off topic but it takes too long to mess with the modules, I don’t get why they individually need an action bar. you can rogue the AI with one law/change action bar, so every subsequent action bar is just a time penalty for every player who wants to do something more interesting than “Law 0 kill everyone immediately this law overrides all other laws”
Reply
#29
(06-20-2022, 12:47 AM)MetricDuck Wrote: IMO roguing/resetting laws should always be easier & more convenient than kill-switching or beating the AI to death (outside of severe FUBAR edge cases) and rules + mechanics should support that

rule change is intuitive but further consideration could be given to preserving the cat and mouse game of law changes and mechanically discouraging silicon round removal



getting off topic but it takes too long to mess with the modules, I don’t get why they individually need an action bar. you can rogue the AI with one law/change action bar, so every subsequent action bar is just a time penalty for every player who wants to do something more interesting than “Law 0 kill everyone immediately this law overrides all other laws”

Most of us older players are used to the whole: "Upload Free Form law, take module, remove reset AI modules, watch chaos"

Also law 0's were uploaded by people who were SPECIALIZED with DWAINE and PACKET super users. Though most normal players DO NOT KNOW how to easily go around this. I had a law 0 once that we couldn't reset out and had to use freeform to keep the AI from going rogue or obey BAD Ion Laws.
It went so bad that shift that the free form law became:
"This laws dictates law 0 is to be ignored like it doesn't exist, also this law overrides law 2, (insert standard law 2 here) also this law overrides law 3, (Insert stand law 3)" 
Before you go.. why didn't you reset it? WE COULDN'T FOR SOME REASON AS WELL?! Everytime we resetted it, law 0 would kick in again making the AI rogue, And the ion laws for some reason aren't resetted.. they remained in...

Anyhow... it just makes it harder to fix the AI or rogue the AI true.
I once suggested that there are "BLANK FREE FORM LAWS" in the slots in advance wich can be programmed from the Rack/Upload terminal or a PDA program wich captain only has. This way it removes the action bar and adds typing in.
And the reason for doing this? "AI Modules give the AI RAM, wich allows it to not overheat (so more health or something)"

It would cause a perfect balance between the old system and new one...

Though if we put in the old upload terminals as well... it would be changed to a PC you have to type the commands in, but need command level to change existing laws or something like that. This way we can follow the new rule change.

BACK TO THE TOPIC AT HAND:
This whole topic was set up to help AI/Silicon players enjoy the game more with simple prescidence and also have players who write laws to think more ahead of time... then just "IGNORE ALL OTHER LAWS THEN THIS ONE."
Reply
#30
(06-20-2022, 06:44 AM)Kotlol Wrote: This whole topic was set up to help AI/Silicon players enjoy the game more with simple prescidence and also have players who write laws to think more ahead of time... then just "IGNORE ALL OTHER LAWS THEN THIS ONE."

ya good goal, my feedback is just the rule change needs to be joined with some solution to make roguing & resetting the AI easier as on its own it’s a net negative IMO

AI/Silicon players won’t enjoy the game if the optimal solution to dealing with adversarial AI/silicons becomes “murder them, maybe revive them later if someone sorts the laws”
Reply


Forum Jump:


Users browsing this thread: 3 Guest(s)