Post by account_disabled on Mar 9, 2024 8:33:41 GMT
Prof Nicola SEARLE When we talk about the creative industries because a lot of AI is text-based at the moment, this is about creations and creative content. If we look back to 20-25 years ago to the advent of the digital economy, how it evolved specifically looking at the music industry, we had Napster where you had no idea who was consuming what, and there was a lot of infringement. Then we moved to iTunes, then later to Spotify, and now TikTok, which is doing well in terms of music, and now we are not quite sure what happens next because there are still tensions. It has been an evolution and I think that we are in the same spot now with AI as we were in the Napster phase. We have a lot of free content, which we are consuming and not paying for these services. We are going to see more advanced tools and the equivalent of iTunes is going to be around the corner, provided that there will be financially viable services for consumers and B2B. So, we are at this point where we don't know what we are dealing with yet, but a lot of copyright questions come up – who owns it, who should be paying for it, who knows what goes in and out, who is responsible.
These are the big questions that the lawyers must grapple with. This should also be analyzed in terms of economic impact. What we see right now is everyone struggling in all these jurisdictions trying to figure out what to do and several lawsuits are going on. It will be interesting to see how this develops and compared to Napster which had no money, now AI is owned by companies with a lot of money, could this UK Phone Number be a Google Books situation? I do not think that at this point we can take sides, it is so unclear what needs to happen or what should happen, and what is fair. A lot of things are happening with copyright and the trade secrets side of it is also very interesting. Samuel LEE: In terms of some background about copyright, the AI engines with slight modifications are pretty much public and certain companies like Open AI or Google their AI engines are slightly different, but in essence, these engines are modeled after language models. As a result, they are trying to predict the next word or the proper sequence of sentences.
What you do is you create this model, and you teach it, feed it, using the computer power that you have with information from tons of books and images, etc. which are copyrighted material. The first question becomes: when you train an AI model, does using this info itself infringe the copyright law or the legal rights that you have? That's a question, because is it reading it? In copyright, one of the rights you have is the right to reproduce, along with other certain rights. If we look at this in terms of a book or music in Napster, if you have a music mp3 file and you download it, get a copy of it, and enjoy that music. In this case, when you use it in the AI model to teach it there is still a question. I think there is an infringement, but not a traditional infringement. These litigations are currently happening with copyright owners of books, images and music against Open AI and other companies such as Google. These questions have not been answered right now and are currently being considered in the legal system. If you are going to use data without any ramifications, this should not be fair. Looking at how Google News service operates, before they would reproduce entire articles, they now put a link to the original news service, and this only occurred because of litigation. I agree with Nicola that issues should be worked on through the courts based on current and future laws depending on which country we are talking about.
These are the big questions that the lawyers must grapple with. This should also be analyzed in terms of economic impact. What we see right now is everyone struggling in all these jurisdictions trying to figure out what to do and several lawsuits are going on. It will be interesting to see how this develops and compared to Napster which had no money, now AI is owned by companies with a lot of money, could this UK Phone Number be a Google Books situation? I do not think that at this point we can take sides, it is so unclear what needs to happen or what should happen, and what is fair. A lot of things are happening with copyright and the trade secrets side of it is also very interesting. Samuel LEE: In terms of some background about copyright, the AI engines with slight modifications are pretty much public and certain companies like Open AI or Google their AI engines are slightly different, but in essence, these engines are modeled after language models. As a result, they are trying to predict the next word or the proper sequence of sentences.
What you do is you create this model, and you teach it, feed it, using the computer power that you have with information from tons of books and images, etc. which are copyrighted material. The first question becomes: when you train an AI model, does using this info itself infringe the copyright law or the legal rights that you have? That's a question, because is it reading it? In copyright, one of the rights you have is the right to reproduce, along with other certain rights. If we look at this in terms of a book or music in Napster, if you have a music mp3 file and you download it, get a copy of it, and enjoy that music. In this case, when you use it in the AI model to teach it there is still a question. I think there is an infringement, but not a traditional infringement. These litigations are currently happening with copyright owners of books, images and music against Open AI and other companies such as Google. These questions have not been answered right now and are currently being considered in the legal system. If you are going to use data without any ramifications, this should not be fair. Looking at how Google News service operates, before they would reproduce entire articles, they now put a link to the original news service, and this only occurred because of litigation. I agree with Nicola that issues should be worked on through the courts based on current and future laws depending on which country we are talking about.