Artisanslop

[Artisanslop @Artisanslostqc]

@BGMaxie @loxeigh @FulIden @benmcoleman @Rahll You don’t have a functional brain, that’s why you can’t understand nuance, the nuance that we’re talking about AI training being theft - which AI hasn’t lost on any such basis. All your articles, for the reasons I mentioned, didn’t prove otherwise. Happy to restate them for you.

[Artisanslop @Artisanslostqc]

@verypeached @DrHasMo @BassFromThePast You’re the one who mentioned value

You explicitly said it’s not valuable, yet it’s solving math equations and problems no one could.

[helios @verypeached]

@DrHasMo @BassFromThePast both the original will smith spaghetti and "evolved" spaghetti have almost nothing of value. at least the first iteration was a meme. also, gen ai can't create anything novel like humans can, since it's not intelligent. so i don't see any future value either.

[Artisanslop @Artisanslostqc]

@verypeached @DrHasMo @BassFromThePast I’m not moving the goalposts, you explicitly talked about “it has no value”. If we’re using that definition of intelligence, then that definition is irrelevant to whether it brings value - because clearly solving math problems nobody has is very valuable.

[Artisanslop @Artisanslostqc]

If you choose to be intentionally dense that’s mostly your issue;

1) No - the court explicitly OKAY’d training on copyrighted sources
2) No - Its a win because the court explicitly okay’s using copyrighted books for training
3) This was not about an LLM like ChatGPT or Claude, so it’s it even relevant
4) This isn’t about training

You should really use AI because you have a very child like or surface level understanding about why you linked. How old are you?

[Artisanslop @Artisanslostqc]

1) Your entire argument was that AI training itself is theft, and that case did not establish that. The court recognized that the TDM rules can apply to AI training in principle. OpenAI lost because specific lyrics were found to be memorised in the model and reproducible in outputs. That is a much narrower issue than “training on copyrighted works is illegal.”

2) Anthropic did not lose on the training question. The court held that training Claude on lawfully acquired books was fair use. The settlement was over the separate class of books Anthropic obtained from pirate libraries and kept. A massive settlement does not somehow reverse the court’s fair use ruling on training.

3) ROSS was not even an LLM. It was a legal-search system using copyrighted Westlaw headnotes to build another competing legal-search product. Fair use failed because it served essentially the same purpose and market. The court even noted that “legal opinions are freely available” and that ROSS could have created what it needed itself from those sources. That is not a ruling that AI training generally cannot be fair use.

4) This has has nothing to do with training infringement at all. You obviously didn’t read the article. It asks whether an AI itself can be a copyright author. The answer was no because copyright requires human authorship.

I say this sincerely, you’d be better off reading Ai generated summaries because you’ve clearly not read or understood what you linked.

[Artisanslop @Artisanslostqc]

@nexusdoll For you? You’ll have to live with the fact your worldview is nonsense when you inevitable won’t be able to defend it. So maybe for your sake you shouldn’t.

[Artisanslop @Artisanslostqc]

@BGMaxie @FulIden @loxeigh @benmcoleman @Rahll https://t.co/v0oNCYMQGC

[Artisanslop @Artisanslostqc]

Unbelievable, you call us out for using AI and “hallucinating” but you clearly hastily googled up articles you obviously didn’t read. I’ll go through one by one

For the first case, the court expressly said that AI models are “in principle” covered by the “text and data mining” rules and that those rules apply to AI training, including necessary copies made to compile and prepare the training corpus. OpenAI lost because the court found these particular lyrics were memorised in the finished model and reproducible in outputs, which it treated as going beyond the TDM exception. So again, the ruling was not that training on copyrighted works is illegal or stealing, the court said TDM can cover AI training, but not retaining recoverable copies of the protected work in the model.

The second article is literally proving OUR point, Anthropic won the case as the court said the actual model training was fair use (not theft). The problem was the separate pirated book library they kept - but most things on the net aren’t behind paywall, so no need to pirate.

The third article isn’t even about an LLM. It was a legal-search product using copyrighted Westlaw headnotes to build another legal search product competing in the same market. The court literally pointed out that “legal opinions are freely available” and that there was nothing Thomson Reuters created that ROSS couldn’t have created for itself. So, it the court said it could have built its training material from the underlying public-domain case law instead of copying Westlaw’s copyrighted editorial summaries. That is very different from a ruling that AI training itself is illegal.

And the last source isn’t about training at all. It’s about whether AI itself can own copyright. It can’t, because copyright requires human authorship.

Ironically claiming the the AI’s are “hallucinating” lands a lot less well when your own sources don’t support the claim you’re making.

[Artisanslop @Artisanslostqc]

Unbelievable, you call us out for using AI and “hallucinating” but you clearly hastily googled up articles you obviously didn’t read. I’ll go through one by one

For the first case, the court expressly said that AI models are “in principle” covered by the “text and data mining” rules and that those rules apply to AI training, including necessary copies made to compile and prepare the training corpus. OpenAI lost because the court found these particular lyrics were memorised in the finished model and reproducible in outputs, which it treated as going beyond the TDM exception. So again, the ruling was not that training on copyrighted works is illegal or stealing, the court said TDM can cover AI training, but not retaining recoverable copies of the protected work in the model.

The second article is literally proving OUR point, Anthropic won the case as the court said the actual model training was fair use (not theft). The problem was the separate pirated book library they kept - but most things on the net aren’t behind paywall, so no need to pirate.

The third article isn’t even about an LLM. It was a legal-search product using copyrighted Westlaw headnotes to build another legal search product competing in the same market. The court literally pointed out that “legal opinions are freely available” and that there was nothing Thomson Reuters created that ROSS couldn’t have created for itself. So, it the court said it could have built its training material from the underlying public-domain case law instead of copying Westlaw’s copyrighted editorial summaries. That is very different from a ruling that AI training itself is illegal.

And the last source isn’t about training at all. It’s about whether AI itself can own copyright. It can’t, because copyright requires human authorship.

Ironically claiming the the AI’s are “hallucinating” lands a lot less well when your own sources don’t support the claim you’re making.

[Artisanslop @Artisanslostqc]

@BGMaxie @loxeigh @FulIden @benmcoleman @Rahll As the guy above showed, they’ve won every case on copyright so far. Do you know why? Because you don’t need “consent” for FAIR USE of copyrighted images on a public domain.

Learn how copyright works before making stubborn opinions.

[Artisanslop @Artisanslostqc]

@BGMaxie @loxeigh @FulIden @benmcoleman @Rahll Now refer to mine where I say how you’re wrong

[Artisanslop @Artisanslostqc]

@BGMaxie @FulIden @benmcoleman @loxeigh @Rahll Except it’s not theft and I guarentee your reasoning about “taking people’s work” misunderstands the fact they train on that work, not generate replicas - thereby not violating any copyright laws under fair use AND the fact all the work is on the public domain.

[Artisanslop @Artisanslostqc]

@TheoTheFurry_ @bodekbodzinski @Rahll @loxeigh @benmcoleman Go study statistics after saying something like this ahaha

[Artisanslop @Artisanslostqc]

@nexusdoll Will you actually argue against my points now or?

[Artisanslop @Artisanslostqc]

@Solstice9684 @Thememe26920328 @TheMG3D Yes they do work, but they don’t render the millions of light rays.

[Artisanslop @Artisanslostqc]

@Panpacificpinez @Thememe26920328 @TheMG3D By comparing ai to commissioning a human, you’re implicitly admitting you see those two as equal.

[Artisanslop @Artisanslostqc]

@verypeached @DrHasMo @BassFromThePast Why are you talking about the value of ai art in a post about coding?

Also the fact you’re reducing solving a math problem NO one Could prior to AI, shows me you’re clearly biased. If it was just “calculator” then everyone should have solved them, and yet.

[Artisanslop @Artisanslostqc]

@BGMaxie @loxeigh @FulIden @benmcoleman @Rahll Thats me doing that, you were talking about “AI” being theft? The reason I did that was to make a point that people will call something bad because they think it’s ai, even though it’s not - which several people ended up doing.!

[Artisanslop @Artisanslostqc]

@BGMaxie @FulIden @benmcoleman @loxeigh @Rahll Except it’s not theft and I guarentee your reasoning about “taking people’s work” misunderstands the fact they train on that work, not generate replicas - thereby not violating any copyright laws under fair use AND the fact all the work is on the public domain.

[Artisanslop @Artisanslostqc]

@Solstice9684 @Thememe26920328 @TheMG3D None of them rendered the millions of light rays manually.

[Artisanslop @Artisanslostqc]

@oscarina0ftime_ @Rahll @benmcoleman @loxeigh And that’s you, everyone else who wants to buy art for far more affordable prices will disagree, anyone else who wants to make their creations come alive will disagree, and so on.

[Artisanslop @Artisanslostqc]

@verypeached @DrHasMo @BassFromThePast If you don’t define that as intelligence, than your definition/word is irrelevant to its value - given solving a math problem is extremely valuable.

https://t.co/0Z6ZxgpQTs

[helios @verypeached]

@DrHasMo @BassFromThePast both the original will smith spaghetti and "evolved" spaghetti have almost nothing of value. at least the first iteration was a meme. also, gen ai can't create anything novel like humans can, since it's not intelligent. so i don't see any future value either.