[Meph504]

> It’s like if I go to Golden Gate Park and pick one flower, I shouldn’t do that, but no one cares. But if I build a machine to automatically cut every flower in the park because I want to sell them, that’s different.

The problem here is, in your example the small scale example, and the large scale example are both unacceptable behavior.

Learning from others at a small scale is not only socially acceptable, but is the foundation of how advancement works.

So this concept of the issue of the scale being the issue isn't at its core the problem, its that something that that is desired behavior in a human, is not socially acceptable because of a machine is doing it.

[customguy]

What a total non-sequitur. You think you found flaw in one of the examples, instead of seeing if you can come up with better ones, you say it so it can't be this, therefore it's a completely different thing that makes zero sense. Machines aren't "doing" anything, they're being wielded by humans. And they're doing it at scale, to other humans, via the force multiplication of machines.

[Meph504]

Why would I attempt to come up with a "better" example of a premise I reject?

Your vague response doesn't seem to have anything to do with the the base subject this whole thing revolves around. Plagiarism be it small scale or large isn't acceptable, and the idea that humans doing things that are wrong is ok, but AI doing the same thing at large scale is not ok?

[customguy]

> Your vague response doesn't seem to have anything to do with the the base subject this whole thing revolves around.

No, I instead refuted your reply.

> but AI doing the same thing at large scale is not ok?

No, humans doing things can be okay or not so okay depending on the scale they do them at. "AI" isn't "doing" anything by itself, at all, so that doesn't enter into it at all. You cannot separate "scale" and "thing". Rubbing your hands to make them warmer is fine, igniting a nuke is not, both aren't "basically the same thing, raising temperature, just at different scales". You didn't reject the premise, you didn't understand it in the first place, and knocked down your own straw man instead. Which I pointed out, that's all.

[Meph504]

Ha, I'm sorry do you think you've made a logical point by comparing rubbing your hands together, and "igniting a nuke."

Again, this isn't a "this at small scale is ok, but at large scale it isn't" argument. Small scale plagiarism isn't acceptable, neither is large scale.

You are refuting my reply seemingly without the context of the article, and larger issue at hand.

Don't be condescending when you aren't even accurately a following the original premise or purpose.

[customguy]

The context of this subthread is explaining that

> If it’s OK (or at least negligible on a small scale), then it must be OK on a large scale.

is a fallacy. Which it is. You confirm this by apparently seeing a difference between generating a little bit of heat and a whole lot, to name one of infinite examples anyone can easily come up with.

> Again, this isn't a "this at small scale is ok, but at large scale it isn't" argument.

You just keep doing the thing I pointed out in my first reply, you claim "it's not this" on a technicality, and then say "so therefore it's this instead", and the other thing is a criticism nobody brings up, ever.

And it gains you nothing, because if plagiarism isn't even okay at small scale, surely you can see how it's even less okay at big scale.

[thedevilslawyer]

There are no such examples (recommended for humans, but abhorrent for machines).

[customguy]

> (recommended for humans, but abhorrent for machines).

That's not the criticism, that's the straw man used to dodge the criticism. Of course the straw man makes no sense, that's why it gets put up.

Machines aren't doing anything, humans are doing things, with or without machines.

"It's fine to raise the temperature of your surroundings by 0.0001 degrees by exhaling. It's less fine to set a house on fire, and even less fine to ignite a nuke. But aren't the all the same thing? How hypocritical that raising temperature is okay for some but not others???"

That things can change quality with quantity/frequency is trivially obvious, and you can think of many examples. Bad ones, good ones, doesn't matter. The point of OP stands, all that was added was how absolutely brazen the nonsense is getting.

[thedevilslawyer]

Sorry, but the point stands for you because of you feel about this topic. It does not stand logically.

Ultimately we have to reckon with the fact that there's nothing which is recommended to do X of, but is abhorrent to do 10X of.

[customguy]

> there's nothing which is recommended to do X of, but is abhorrent to do 10X of.

No we don't, because that's nonsense. You can ask a stranger in the street for the time of day once, and they will react very, very differently if you ask them 10 times in a row. You can drive N miles per hour in a school zone, you cannot drive at 10x the speed, and so on.

[thedevilslawyer]

Ok, I see your point. We live with tiny inconveniences, that we would not at 10x.

But I don't see how that relates to copyright or llm at all. 'Learning', at scale, is not an inconvenience, atleast in any forward looking society.

[musicale]

> There are no such examples (recommended for humans, but abhorrent for machines)

claiming to be human

[dinfinity]

> Learning from others at a small scale is not only socially acceptable, but is the foundation of how advancement works.

Exactly, if anything, the logic (a bit bad -> really bad) shows that one person learning from one thing is far inferior to one person learning from every thing (a bit good -> really good).

[insane_dreamer]

> The problem here is, in your example the small scale example, and the large scale example are both unacceptable behavior.

no, not really, or at the very least they're not at all in the same category of "unacceptable behavior"

[Meph504]

The argument isn't small crime vs large crime. It is no crime regardless of scale.

If it is acceptable for a person to learn, then it should be acceptable for a machine. And any derived works produced from that information isn't theft or copyright violation.

Though I do think there is a valid gripe with the LLMs being trained on pirated materials. I've also personally learned from a lot of PDF of textbooks I didn't own.

[dtn]

> If it is acceptable for a person to learn, then it should be acceptable for a machine.

Is there a name for the fallacy when people act like models and algorithms should be granted the same rights as human beings?

[simianparrot]

AI psychosis would be my term. When you attribute to software the characteristics of a human you are in a psychosis.

[Meph504]

Many things share characteristics with human, we have for decades created methods for systems to emulate and synthesize those characteristics. It is sort of delusional to think that the abilities of humans can't be produced by other systems, it is a severe delusion to think that proposing a machine can do it, is psychosis.

[simianparrot]

That’s not the topic. Read the post I replied to. No matter what a piece of software will never be a human.

[Meph504]

Is it a fallacy? Can you provide a legal or logical basis for this being treated differently.

[lelanthran]

Hammers aren't granted rights.

Tools aren't granted rights. Why do we need to make an exemption for AI?

[Meph504]

Well because no one is attempting to claim that the structures or products produced by using Hammers are plagiarism?

In a sane world, things produced by tools are owned and credited as creations by the users of tools, there are many who seem to argue that isn't the case with AI.

And that some how, that anything produced based on the knowledge it was trained on is some sort of plagiarism or copyright violation of the original source material even when none of that material is present in the end result?

So if we can't just leave it at its a tool, then we have to look at existing frameworks of laws and ethics to make the case of how this should be treated.

[mark336]

>Learning from others at a small scale is not only socially acceptable, but is the foundation of how advancement works.

This is true, shows how human thought differs from AI. AI needs massive datasets to be coherent.

[Meph504]

How large do you think your some total of all things learned would be as a dataset, we aren't that different in that regard, just in how we amass that dataset and how curated it is.

[LunicLynx]

Wasn’t his point about plagiarism? That is also not ok on a small scale.

[Meph504]

I was trying to stick to the example, but I agree, that getting away with something doesn't determine if it is right or wrong. And the whole concept of that makes for shaky ground for any form of legal or ethical argument.

[boringg]

I think the difference here is that you guys are talking ethics. And in fact what were talking about is enforcement. While its unethical to pick one flower (in it's purest form, robbing the commons of the beauty of a flower), it won't be enforced.

[LunicLynx]

Fair. AI might also not be the problem, but how it is utilized.

Suddenly everyone and their grandma are specialist at everything and the actual value of understanding is not appreciated anymore.

[thedevilslawyer]

Ok, what is so special about understanding anyway? we understand way less things than we do no understand.

IMO, we're just giving special weight to understanding just because it gives people wages. Someone's specific brain structure should not privilege them over others. UBI or something equitable on those lines is the answer.