[DRAFT] LAWS Monitoring Act

[Araraukar]

KNOWING that one of the advancements a nation can make in technology is Artificial Intelligence, and that technologies such as this are used in Lethal Autonomous Weapon Systems (LAWSs).

OOC: This needs to be in the active clauses as a definition, not in the preamble. Also, artificial intelligence refers to human-grade-or-higher intelligence, so think of a different wording, given you're specifically not talking about those.

RECOGNIZING the immorality embedded in giving a machine full responsibility over life and death situations on the battlefield.

This is also irrelevant, given you're not doing that.

BELIEVING that there should always be some degree of sapient intervention in offensive weapons systems.

Hence mentioning AIs in the first preamble clause is a bad idea and confuses everyone.

RESOLVING that there should be international monitoring and restrictions set on LAWSs.

Why?

On this matter, the GA thus declares that, except as limited by earlier resolutions that are still in place:

Start the whole proposal with "The World Assembly" or "The General Assembly", and replace this line with "Hereby, [excemption goes here]".

LAWSs are defined as weapons systems that operate independently from their sapient users while being used and are programmed to engage in lethal tactics.

Spell out the full words - basically move the thing from preamble to here. Also, if they're fully automated, they don't really have "users" as such, now do they? Deployers, maybe?

The World Assembly mandates boycotting the manufacturing of and use of offensive LAWSs on the battlefield to all of its members.

You still haven't explained what makes them so bad that they would need to be banned. Also, if you're allowing their use defensively, why are you requiring "boycotting the manufacturing of"? And why boycotting? Why not just banning?

The World Assembly mandates that to be manufactured, sold, bought, transferred and used by its members, a LAWS must have the following qualities:

1. Can explain their reasoning and decisions to sapient operators in transparent and understandable ways.
2. Have responsible sapient operators who are clearly identifiable.
3. Have autonomous functions that are predictable to their operators.
4. Be manufactured for defensive purposes and used solely by defensive units.
5. Fit into previous laws of humane weaponry.

But their manufacture must still be boycotted? Also, if you've got a battlefield, and you put an "automated machine gun" (your choice of example!) there to defend your own people, and move it forward as you advance, is it still defensive use? And do aerial drones carrying missiles count? You say in the thread that guided missiles don't count, but I'm not seeing anything here that would make them not count. Especially as missiles are hardly either humane or defensive use items.

[Kenmoria]

“Clause 3 is rather broad in terms of banning a LAWS from being ‘manufactured, sold, bought, transferred and used’ without any quantifiers. What about manufacturing one for educational reasons to show what isn’t allowed? How about transferring a LAWS that breaks the rules to a waste disposal facility? What of transferring a LAWS into a non-member nation where it may be used?”

[Danlina]

Will it fix the problem if I abandon the word “transferred”?

[Kenmoria]

Will it fix the problem if I abandon the word “transferred”?

(OOC: I think restricting it to only banning the usage, as opposed to anything else, of LAWSs would fix it, since one can’t really do any damage with a LAWS without its usage.)

[Araraukar]

OOC: And please remove or spoiler the massive picture from the first post.

[Marxist Germany]

OOC: And please remove or spoiler the massive picture from the first post.

OOC:Or resize it if possible to a size that is minuscule.

[Widowed Land]

I’m just trying out a winning MUN legislation of mine to see if this game is even suitable for practicing diplomacy.

OOC: hah ^_^ I did the same(not 100% copy of MUN resolution, but some clauses were copied) But nobody liked it so I had to rewrite it...

[Kenmoria]

“I find your ‘recognising’ clause to have rather meandering logic. What exactly is so bad about giving a machine control over life and death decisions? Surely it is better than embodying the responsibility in something fallible? The clause also ignores the multitude of AI nations.”

[Danlina]

So I tweaked the wording and added Danlina’s reasoning to believe offensive LAWSs should be banned. Any other arguments or problems I should take into account?

[Araraukar]

Any other arguments or problems I should take into account?

OOC: You're still mentioning AIs as being a problem, despite what you want to legislate having nothing to do with AIs or even machine learning.

[Danlina]

Any other arguments or problems I should take into account?

OOC: You're still mentioning AIs as being a problem, despite what you want to legislate having nothing to do with AIs or even machine learning.

Where did I say that AIs themselves are the problem? I said that the use of non-sentient AIs in weapons systems is the problem.

[Kenmoria]

“Also, please put line breaks between your active clauses, to aid clarity.”

[Araraukar]

I said that the use of non-sentient AIs in weapons systems is the problem.

OOC: Artificial Intelligence =/= non-sentient, even the way it's understood in Real Life.

Also your definitions don't fit the way that "artificial intelligence" (which encompasses sentience and machine learning, which means the program(s) take data from the external stimuli and react to it independently, and learn to get better at it) is usually meant. I think yours would be closer to "smart programs" (which are only smart the way smartphones are smarter than normal phones). And the way you define the "allowed systems" later in the proposal, they don't really fit your own definition.

[Danlina]

I said that the use of non-sentient AIs in weapons systems is the problem.

OOC: Artificial Intelligence =/= non-sentient, even the way it's understood in Real Life.

Also your definitions don't fit the way that "artificial intelligence" (which encompasses sentience and machine learning, which means the program(s) take data from the external stimuli and react to it independently, and learn to get better at it) is usually meant. I think yours would be closer to "smart programs" (which are only smart the way smartphones are smarter than normal phones). And the way you define the "allowed systems" later in the proposal, they don't really fit your own definition.

I am sorry, I thought it was clear what a sentient AI is in contrast to a MT AI: Sentience in a machine is determined by the presence of self-awareness and if it passes the Turing test. In other words, in MT and usually even PMT sentient AI is non-existent, and in FT it exists alongside non-sentient AI. This means it is possible to ban LAWSs without hurting any national laws that give sentient machines equal rights because the technology used in LAWSs isn't sentient.

[The First German Order]

“The Order is completely against this proposal. We do not use unmanned but operated combat units on the battlefield, but rather we use unmanned semi-sentient combat robots. As such, this proposal in its current state would require that we either risk using human combatants or spend possibly billions to modify ALL of our current combat automatons to be controlled by operators AND enlist who knows how many individuals to operate said robots. And, contrary to what was previously stated in response to a fellow representative, our robots are prefectly capable of identifying a surrendered combatant.”

OOC: Artificial Intelligence =/= non-sentient, even the way it's understood in Real Life.

Also your definitions don't fit the way that "artificial intelligence" (which encompasses sentience and machine learning, which means the program(s) take data from the external stimuli and react to it independently, and learn to get better at it) is usually meant. I think yours would be closer to "smart programs" (which are only smart the way smartphones are smarter than normal phones). And the way you define the "allowed systems" later in the proposal, they don't really fit your own definition.

I am sorry, I thought it was clear what a sentient AI is in contrast to a MT AI: Sentience in a machine is determined by the presence of self-awareness and if it passes the Turing test. In other words, in MT and usually even PMT sentient AI is non-existent, and in FT it exists alongside non-sentient AI. This means it is possible to ban LAWSs without hurting any national laws that give sentient machines equal rights because the technology used in LAWSs isn't sentient.

OOC: Uh. My sir, I’m pretty sure that the Turing Test is for sapience, not sentience.

[Araraukar]

OOC: Uh. My sir, I’m pretty sure that the Turing Test is for sapience, not sentience.

OOC: ^This.

I know that outside of specific contexts like this, "sentience" and "sapience" are used as synonyms, but it's incorrect usage (even outside of here).

In Real Life a regular cat or dog is sentient. As is a salmon. As, arguably, is a daisy as well. Or a butterfly.

In Real Life, humans and other apes, dolphins, elephants and some birds are sapient in addition to being sentient.

In Real Life, "sentience" is not used about artificial intelligence, whether or not talking about existing sentient programs or about hypothetical sapient ones.

In Real Life, machine learning is a complex thing, but there already are programs that would fit the sentience definitions, and robots using them, that definitely exhibit sentience.

So while the proposal's current language excludes sapient AIs, any kind of automated weapons system would pretty much by definition be sentient. Otherwise how could it react to target things and shoot at them?

Also, this is still unaddressed:

"By removing offensive automated weapons, you force manned weapons to take their place, which increases the risk to life and limb. That is most unwanted, especially since the underlying premise of your proposal makes indirect fire from artillery just as immoral as a hunter-killer drone. And that is assuming that morality should have anything to do with it."

As are these:

So I'm really struggling to understand how the "can explain their reasoning and decisions" wouldn't fall under sapient AIs, unless "reasoning" has a different meaning from the usual one.

Again, you're using "explain their reasoning and decisions" in a way that the words are not normally meant. Spell out your actual meaning rather than making it appear the weapons need to be able to talk or be sapient themselves.