[ukFxqnLa2sBSBf6]

I consider myself progressive and my main issue with the technology is that it was created by stealing from people who have not been compensated in any way.

I wouldn’t blame any artist that is fundamentally against this tech in every way. Good for them.

[notJim]

Every artist and creator of anything learned by engaging with other people's work. I see training AI as basically the same thing. Instead of training an organic mind, it's just training a neural network. If it reproduces works that are too similar to the original, that's obviously an issue, but that's the same as human artists.

[_DeadFred_]

Human beings are human beings.

For profit products are for profit products, that are required to compensate if they are derivative of other works (in this case, there would be no AI product without the upstream training data, which checks the flag that it's derivative).

If you would like to change the laws, ok. But simply breaking them and saying 'but the machine is like a person' is still... just breaking the laws and stealing.

[dweinus]

This is a bad-faith argument, but even if I were to indulge it: human artists can/do get sued for mimicing the works of others for profit, which AI precisely does. Secondly, many of the works in question have explicit copyright terms that prohibit derivative works. They have built a multi-billion dollar industry on scaled theft. I don't see a more charitable interpretation.

[notJim]

You can't call something a bad-faith argument just because you disagree with it. I mean, you can, but it's not at all convincing.

As I said, if AI companies reproduce copyrighted works, they should be sued, just like a human artist would be. I haven't experienced that in my interactions with LLMs, but I've never really tried to achieve that result either. I don't really pirate anymore, but torrents are a much easier and cheaper way to do copyright infringement than using an AI tool.

[blackqueeriroh]

LLMs don’t have to be able to mimic things. And go ahead and sue OpenAI and Anthropic! It won’t bother me at all. Fleece those guys. Take their money. It won’t stop LLMs, even if we bankrupted OpenAI and Anthropic.

[hackable_sand]

> I see training AI as basically the same thing

Of course you do.

[musicale]

It's "unauthorized use" rather than "stealing", since the original work is not moved anywhere. It's more like using your creative work to train a software system that generates similar-looking, competing works, for pennies, at industrial scale and speed.

[andrei_says_]

Obtaining without payment or consent and then using to create derivative works at scale?

And the pedantry matters only because the entities criming are too big and rich and financed by the right people.

It is basically a display of the societal threshold beyond which laws are not enforced.

[musicale]

> Obtaining without payment or consent

Usually "obtaining" is just making a bunch of HTTP requests - which is kind of how the web is designed to work. The "consent" (and perhaps "desired payment" when there is no paywall) issue is the important bit and ultimately boils down to the use case. Is it a human viewing the page, a search engine updating its index, or OpenAI collecting data for training? It is annoying when things like robots.txt are simply ignored, even if they are not legally or technically binding.

The legal situation is unsurprisingly murky at the moment. Copyright law was designed for a different use case, and might not be the right tool or regulatory framework to address GenAI.

But as I think you are suggesting, it may be an example of regulatory entrepreneurship, where (AI) companies try to move forward quickly before laws and regulations catch up with them, while simultaneously trying to influence new laws and regulations in their favor.

[Copyright law itself also has many peculiarities, for example not applying to recipes, game rules, or fashion designs (hence fast fashion, knockoffs, etc.) Does it, or should it, apply to AI training and GenAI services? Time will tell.]

[crimsoneer]

The pedantry matters for the same reason it mattered when the music industry did this to Napster: because the truth is important.

[musicale]

Accurate terminology can help to clarify discussions, including discussions of appropriate activity related to creative works.

My suspicion is also that what many artists are objecting to is less the training itself (though AI web crawlers are aggressive and annoying) and more the use of the trained model for large-scale generation of similar (and possibly competing) works ("derivative" in an artistic if not legal sense), in the artist's style, especially by large companies for commercial gain.

[jimbokun]

Ok Mr. (Or Ms.) Pedant you know what the intended meaning was.

[taco_emoji]

everybody knows this, you are being uselessly pedantic

[blackqueeriroh]

Many of those people have absolutely been compensated, many times over. Should they have a perpetual right to longer than their entire life, to profit off something they did 40 years ago?