Technology

Is the ai doing anything that isn’t already allowed for humans. The thing is, generative ai doesn’t copy someone’s art. It’s more akin to learning from someone’s art and creating you own art with that influence. Given that we want to continue allowing hunans access to art for learning, what’s the logical difference to an ai doing the same?

AI stans always say stuff like this, but it doesn't make sense to me at all.

AI does not learn the same way that a human does: it has no senses of its own with which to observe the world or art, it has no lived experiences, it has no agency, preferences or subjectivity, and it has no real intelligence with which to interpret or understand the work that it is copying from. AI is simply a matrix of weights that has arbitrary data superimposed on it by people and companies.

Are you an artist or a creative person?

If you are then you must know that the things you create are certainly indirectly influenced by SOME of the things that you have experienced (be it walking around on a sunny day, your favorite scene from your favorite movie, the lyrics of a song, etc.), AS WELL AS your own unique and creative persona, your own ideas, your own philosophy, and your own personal development.

Look at how an artist creates a painting and compare it to how generative AI creates a painting. Similarly, look at how artists train and learn their craft and compare it to how generative AI models are trained. It's an apples-to-oranges comparison. Outside of the marketing labels of "artificial intelligence" and "machine learning", it's nothing like real intelligence or learning at all.

(And that's still ignoring the obvious corporate element and the four pillars of fair use consideration (US law, not UK, mind you). For example, the potential market effects of generating an automated system which uses people's artwork to directly compete against them.)

You picked the wrong thread for a nuanced question on a controversial topic.

But it seems the UK indeed has laws for this already if the article is to believed, as they don't currently allow AI companies to train on copyrighted material (As per the article). As far as I know, in some other jurisdictions, a normal person would absolutely be allowed to pull a bunch of publicly available information, learn from it, and decide to make something new based on objective information that can be found within. And generally, that's the rationale AI companies used as well, seeing as there have been landmark cases ruled in the past to not be copyright infringement with wide acceptance for computers analyzing copyrighted information, such as against Google, for indexing copyrighted material in their search results. But perhaps an adjacent ruling was never accepted in the UK (which does seem strange, as Google does operate there). But laws are messy, and perhaps there is an exception somewhere, and I'm certainly not an expert on UK law.

But people sadly don't really come into this thread to discuss the actual details, they just see a headline that invokes a feeling of "AI Bad", and so you coming in here with a reasonable question makes you a target. I wholly expect to be downvoted as well.