Finally, AI Lawsuit Filed by Grisham, Martin, and Hilderbrand

1. 92

   Kenna McHughposted 7 months ago

   As a freelance writer, the noose around my neck is getting tighter with AI content expanding. I am happy to see this happening.
   OpenAI, the creator of ChatGPT, is facing a lawsuit from bestselling writers, including George R.R. Martin, John Grisham and Elin Hilderbrand, that claims the company fed their books into its "large language models" allegedly violating their copyrights and engaging in "systematic theft on a mass scale."
   
   https://www.cbsnews.com/news/openai-law … =236407565

   replyreport
   1. 92

      jansharesposted 7 months agoin reply to this

      Wow. This is huge, precedent setter for sure.

      replyreport
      1. 92

         Kenna McHughposted 7 months agoin reply to this

         Jan, Yes! It will get attorneys, telling them to slow down, check your programs.

         replyreport
   2. 89

      Nell Roseposted 7 months agoin reply to this

      Thank goodness for that. I tried ai writing out last week, just to see the hype. Then put the results on Facebook. Pretty scary to be honest. I wrote one sentence then generated it. It came up with a near-perfect scenario. Facebook readers said it was generic, so no worries,  but still, quite annoying and scary.

      replyreport
   3. 90

      Patty Inglish, MSposted 7 months agoin reply to this

      I hope the lawsuit is successful. Harlan Ellison would have had one of his cursing fits on YouTube about AI content, if he were still alive. I'm glad he missed AI and its harm to writers.

      replyreport
   4. 68

      Miebakagh57posted 7 months agoin reply to this

      OMG chap gpt a copy write theft?

      replyreport
   5. 92

      Jan Stepanposted 7 months agoin reply to this

      It raises awareness about the issue and creates discussion, which is positive, but I sadly think it won't change anything.
      
      They are chanceless, and I think deep down they know it themselves. I hope I am wrong on this one.

      replyreport
      1. 92

         Kenna McHughposted 7 months agoin reply to this

         Certain institutions, such as the Writers and Illustrators of the Future contest, forbid AI. They do not allow any AI.

         replyreport
2. 43

   Palmershadyposted 7 months ago

   This is great news!

   replyreport
   1. 92

      Kenna McHughposted 7 months agoin reply to this

      Yes. It is. I wonder how it will play out.

      replyreport
3. 96

   Rupert Taylorposted 7 months ago

   I wish them well. The writers quoted have deep pockets and it may cost them a lot to get a judgement against ChatGP, which will probably immediately declare bankruptcy and the plaintiffs will get nothing.
   
   Then, a different scumbag vulture will step in to profit off the work of others.
   
   Sorry, but I have a very jaundiced view of capitalism as presently constituted.

   replyreport
   1. 92

      Kenna McHughposted 7 months agoin reply to this

      Yeah. You're not the only one.

      replyreport
4. 96

   Thomas Swanposted 7 months ago

   I don't see how it would succeed. ChatGPT would argue that human brains work in the same way: feeding in books and writing new stuff based on what we've fed on.

   replyreport
   1. 92

      Kenna McHughposted 7 months agoin reply to this

      I hear you. However, there is copyright law. Writers can't take information from a book verbatim without proper references. AI needs to do the same. Even spinning it, which AI does, is a big no-no, especially in academia.

      replyreport
      1. 96

         RonElFranposted 7 months agoin reply to this

         Kenna, I think (based on my admittedly limited understanding of the technology) that the issue will be that a Large Language Model or generative AI like ChatGPT actually doesn't copy or spin the info it ingests. The LLM is based on a huge software-based neural network with perhaps billions or even trillions of nodes (like human brain neurons). Each node, or artificial neuron, is connected to a multitude of others, with each connection path using a specific, adjustable mathematical weighting function that modifies the "signal" traveling along the connection from that node to another.
         
         As the LLM ingests info during training from the web or elsewhere, it doesn't copy the content verbatim. Rather, each item of info it ingests causes some of the weighting factors to be slightly altered. The actual info doesn't exist verbatim anywhere in the model, but the weights allow the model to predict the next word that should statistically occur as it responds to a prompt.
         
         This is, by design, analogous to the way humans learn. During training the text or content isn't copied but only used to adjust artificial neurons. Neither is it directly copied during retrieval, though it might be reproduced (sometimes inaccurately) because of that statistical prediction process. (That's why one of the big issues in AI is that humans usually can't understand how the AI reached its conclusions because there's no chain of reasoning based on stored info).
         
         My take at this point is that if OpenAI and other generative AI producers are able to lucidly explain what the tech actually does, plaintiffs will have a very difficult time proving copyright violations since their work wasn't actually copied and doesn't exist as any coherent body of content within the LLM. If the AI is guilty of copyright violation for learning this way, so is every person who reads a book.

         replyreport
         1. 92

            Kenna McHughposted 7 months agoin reply to this

            Ron, I didn't know that. You're understanding is quite extensive. Thank you.

            replyreport
         2. 97

            theraggededgeposted 7 months agoin reply to this

            You appear to have a great understanding, Ron. I think this lawsuit is doomed to fail because Chat AI does not reproduce or republish copyrighted content.
            
            It is exactly how humankind has progressed... by building on what has gone before. The difference with AI is how fast it is doing so. That's why we are all getting our knickers in a twist.
            
            The way I see it is that it will weed out the mediocre from the innovative. Those of us (and I include myself) who produce work by researching what others have already written need to take a long hard look at what we are doing. And perhaps do it differently.

            replyreport
         3. 96

            Daughter Of Maatposted 7 months agoin reply to this

            I think the problem will be that the AI was allowed to use their copyrighted works for training without permission. Humans have to buy books or borrow them from the library (who purchased them or they were donated by someone who purchased them), so we have the right to read the book and "learn" from it because we paid for the rights to do so. ChatGPT did not.
            
            The question will be, were did the AI get the material? Did the owners actually purchase the books to teach the AI? Or did the AI "scrape" the content from the internet? In which case, I'm sure there are illegal copies of said books on the dark web/general internet, but I would assume the AI used summaries from bloggers pages etc to glean the info. So does ChatGPT have access to the dark web? Or did it simply get the info from summaries all over the internet and because it is an AI it was able to deduce the information based on the summaries it was able to find?
            
            AI is capable of learning much more quickly than we are, depending on how well it was programmed. So it is possible for it to "read" a whole novel from the summaries and discussions online about these books. Not to mention I'm sure there are quotes from the books everywhere. For example, often times Amazon has a chapter for free in the description of the book, so the AI could read just that one chapter and be able to write a whole book in the tone and voice of the original author.
            
            I'm very interested to see how this plays out. If the authors win, it could be the best thing for us, because it may prevent Google (and Bing) from using their AI based search engines to spit out answers to questions that are so thorough the querent won't need to click on any other links, saving us from losing traffic. But it may also stifle the evolution of AI. This is definitely a precedent setter for sure.

            replyreport
            1. 84

               tsmogposted 7 months agoin reply to this

               Just for info, two sources for free books I have used in the past are:
               
               Welcome to Project Gutenberg
               Project Gutenberg is a library of over 70,000 free eBooks
               https://www.gutenberg.org/
               
               and,
               
               A Free Library of Classic Books & Novels
               https://americanliterature.com/books/

               replyreport
            2. 92

               Kenna McHughposted 7 months agoin reply to this

               I want to read the attorneys' briefs because are there precedented cases? Napster? Justice is fickle. A user could ask AI to write a short story in Grisham style. Is that a violation, or do copyright laws need revision?

               replyreport
            3. 96

               RonElFranposted 7 months agoin reply to this

               I think the counter to this argument is that the AI can only scan content that's been made freely available on the web. If you put your content on the open internet you're giving an explicit invitation for people (and computers) to read it for free--otherwise, you put it behind a paywall like the NY Times or Washington Post do for their content.
               
               I'm pretty sure no reputable AI company would be so unwise as to breach a paywall to scan material that is explicitly only available to those who pay a fee. If they did so there's no question they'd lose a copyright suit and would probably be liable for massive damages.

               replyreport
5. 97

   theraggededgeposted 7 months ago

   So I asked Chatty....
   
   Chat AI, like me, learns from books and other text by training on vast amounts of written material. We use a process called unsupervised learning, where we analyze patterns, language structures, and relationships between words and phrases in the text. It's like learning the language by immersion! Then, we can generate text based on what we've learned. It's pretty cool, right?
   
   Me; But do you access copyrighted books?
   
   ChatGPT:
   
   Nope, I don't have access to specific copyrighted books or any proprietary databases. I've been trained on a diverse range of internet text up until my last knowledge update in September 2021. So, I can provide information and generate text based on that, but I can't access or quote from specific copyrighted books. If you have any questions or need information, feel free to ask, and I'll do my best to help!

   replyreport
6. 96

   PaulGoodman67posted 7 months ago

   Quite a few newspapers and publications have "opted out" of ChatGPT.
   
   It's not clear to me what the advantage is for anyone is have AI use their intellectual property to generate money for Bing, Google, or another commercial company.
   
   Unless there's some sort of payment scheme, I don't see how it's going to work.
   
   It's a bit like the early days of digital sampling when music artists could just steal clips from others for free. Eventually, they didn't ban sampling but they did make artists pay for the clips.
   
   AI offers many opportunities but the commercial side will have to be resolved. The new technology renders the old copyright laws inadequate.

   replyreport
   1. 96

      eugbugposted 7 months agoin reply to this

      The thing about AI as Ron suggests though is that AI doesn't sample as such. So like us, it just bases what it creates on the extensive knowledge it has learned.during training. Then it uses that information to write content "in its own words". As regards sampling music, AI can compose, but I guess what is copying is debatable. How many consecutive notes or musical phrases would be needed in a piece of music for a court to decide that the work was copied? It can create content in the style of another author or artist though and that maybe could be considered as copying.

      replyreport
      1. 96

         PaulGoodman67posted 7 months agoin reply to this

         Yes, and that's why the old copyright rules increasingly seem archaic.
         
         AI freely takes and uses intellectual property that others have sweated blood over and perhaps incurred financial costs for.
         
         However, AI's not clearly breaking any of the current copyright rules by doing this (I'm talking generally, not the specific lawsuit in the OP).
         
         There's a game of catch-up likely to happen was my point, or maybe cat and mouse game is a better phrase.
         
         There will likely be turbulence before things settle into a new system. It generally seems bad for writers and creators, though.

         replyreport
         1. 97

            theraggededgeposted 7 months agoin reply to this

            Remember the kerfuffle over Pinterest? It all got a bit blurry around then. Funnily enough, there are tons of AI-generated images on there lately.

            replyreport
            1. 92

               Kenna McHughposted 7 months agoin reply to this

               Even Shutterstock will create AI-generated images. The quality is poor, though it might work if you're not finding what you're looking for on the Internet.

               replyreport
               1. 97

                  theraggededgeposted 7 months agoin reply to this

                  Yeah... what I meant in the first sentence was the controversy over the pinning of copyrighted images.

                  replyreport
7. 96

   RonElFranposted 7 months ago

   I think that as writers we just have to recognize that the AI genie is out of the bottle and it's never going back in.
   
   Even if American courts say that scanning the web to train an AI like ChatGPT violates copyright laws, what would be the practical effect? Given the massive amounts of info required to adequately train a generative AI model, for producers to pay even a single penny to every source they scanned would make the whole process prohibitively expensive. The effect would be to gut the generative AI business and stunt development of the technology in the US.
   
   But that won't stop China and other countries, which would not be bound by US court rulings. They would be ecstatic since it would effectively put the US out of the ballgame and leave the entire playing field to them. Companies that require written content would still have just as much access to AI-generated content, just not from US products.
   
   Generative AI is expected to be as impactful a technology as the internet itself. Do we really want to see the US cede its leadership position in this arena to others when there would be little advantage to writers (or anyone else, for that matter) for doing so?
   
   Yes, we as writers are already being negatively impacted, and that impact will probably grow. But I don't think retarding the development and deployment of generative AI in the US will change that situation for the better. We'll have to find other ways to adjust.

   replyreport
   1. 92

      Kenna McHughposted 7 months agoin reply to this

      Good points, Ron. As Beth said, mediocre writers feel more pain than innovative writers. It's also in the visual art world, though the AI hasn't yet figured out how to create layers. It hurts the newbie artist who is trying to build a career because there are those who will generate mediocre visual pieces for their project instead of giving a newbie a break or a low rung to start their career.

      replyreport
   2. 92

      Jan Stepanposted 7 months agoin reply to this

      Yes. When you pull the trigger, there's quite literally no going back. What we witness here is a rapid development race. If you take a break, you're out of the game. That's why nobody won't and even can't stop. It's a technological rat race. 
      
      Being on top of it means having dominance over the market, and being on top of the market – well, I don't need to teach anyone how much power that brings.

      replyreport
8. 96

   eugbugposted 7 months ago

   For those of us who write guides/tutorials, luckily AI hasn't the ability yet to add specific images relevant to the text. Maybe when it gets around to doing that, it'll use public domain images. Us humans still have the edge though, whereas the AI can only use generic and more general photos.

   replyreport
9. 68

   Miebakagh57posted 7 months ago

   As regards the last paragraph and sentence, this is hard to undetstand.

   replyreport
10. 94

    Readmikenowposted 7 months ago

    This type of copywrite infringement has happened before, but not with sophisticated technology.
    
    At times it has been done quite openly without infringing on copyrighted material.
    
    There was an attempt to write a sequel to "Gone with the Wind," called "Scarlett," which, in my opinion, captured the writing style but not the excellent story development of Margaret Mitchell. I found it a disappointment.
    
    One of my other favorite books of all time is "The Confederacy of Dunces" by John Kennedy Toole. A sequel was written called "A Cornucopia of Dunderheads." Again, it captured the writing style but not the character development or story.
    
    So, can AI do any better than those who try to imitate the writings of other authors? 
    
    Maybe on a larger scale.  Those who are true fans of an author's work can and will always be able to tell the difference.
    
    The problem is the general public may not.

    replyreport
    1. 96

       PaulGoodman67posted 7 months agoin reply to this

       "...can and will always be able to tell the difference"
       
       I don't think that's certain at all. If anything, the opposite is true, I believe. At a certain point, you WON'T be able to tell the difference.
       
       We're just at the start of AI but it's developing at rocket speed!
       
       That doesn't mean that the AI will be conscious. I doubt that it will be able to feel like a human or experience, say, empathy. But it will be effective at producing convincing writing, I believe, particularly in the nonfiction sphere.

       replyreport
11. 96

    Glenn Stokposted 7 months ago

    There’s a lot of controversy about this subject, as I see different viewpoints in the various comments. However, I’m glad to see these lawsuits occurring because ChatGPT has been taking content from our articles verbatim.
    
    “ChatGPT's interaction is pulled from … books, documents, news articles, scientific journals, and the web.” Excerpt from: https://www.readersdigest.ca/home-garde … s-chatgpt/

    replyreport
    1. 92

       Kenna McHughposted 7 months agoin reply to this

       Glenn, Hopefully these lawsuits will glean appropriate information for writers, artists and consumers.

       replyreport
12. 96

    PaulGoodman67posted 7 months ago

    I've been using AI to generate images at Canva for new Pinterest pins. It is amazing what the new technology can do, although there's still room for improvement.
    
    The only way that I can see AI being impeded is at source, that's to say that it's not given access to certain info. Once the unique text or image has been created, I think it's difficult, if not impossible to stop using legislation.

    replyreport
13. 59

    Eva Sharpposted 6 months ago

    I wonder how all of this will end. Moreover, I heard that Google Books essentially used all books to train its robots. Will anyone want to try suing Google?

    replyreport
    1. 92

       Jan Stepanposted 6 months agoin reply to this

       I assume that even if they tried their very best, it still wouldn't change anything.
       
       Finding evidence and, more importantly, showcasing such proof is almost impossible – unless you have a remarkably trustful whistleblower with undeniable access to transparent data who also doesn't care about ruining their career...

       replyreport
       1. 92

          Kenna McHughposted 6 months agoin reply to this

          I would suspect there is a trail of some sort within the system.

          replyreport