5 ms·

There is no proof so far that openAI has violated any copyright. I understand the public’s emotions but recall that these large language models can read all th

• Retric 3y ago
  Being public doesn’t mean copyright goes away. If you can’t use a 3rd party program even once without permission why would you assume you could use someone’s book etc to train a model without similar permission? My guess is there’s going to be an exception given, but that’s not yet decided.
  • tracerbulletx 3y ago
    Because the law has never delt with the concept of training models like these before and it's not just automatically the same thing as republishing something non-transformative no matter how much some people want to just assert that it is without actually thinking about it. It's a new case, that probably needs new different types of protections and consideration.
    • oh_sigh 3y ago
      Does the law care about the method(whether it is ML or just reading the article from a database) if the model can spit out a verbatim copy of an article?
      • MeImCounting 3y ago
        I see this argument everywhere and have yet to succeed in getting ChatGPT to give an article. This argument is basically a non-starter
        • Retric 3y ago
          Researchers have gotten large chunks of verbatim text. Suggesting it’s only infringing if the entire work is reproduced is legally incorrect and therefore irrelevant. https://mashable.com/article/chatgpt-revealed-personal-data-verbatim-text-attack-researchers https://mashable.com/article/chatgpt-revealed-personal-data-...
          • paulddraper 3y ago
            > Suggesting it’s only infringing if the entire work is reproduced is legally incorrect It may not be conclusive, but it's highly suggestive: > Amount and substantiality of the portion used in relation to the copyrighted work as a whole: Under this factor, courts look at both the quantity and quality of the copyrighted material that was used. If the use includes a large portion of the copyrighted work, fair use is less likely to be found; if the use employs only a small amount of copyrighted material, fair use is more likely. > https://www.copyright.gov/fair-use/ https://www.copyright.gov/fair-use/
            • Retric 3y ago
              Context is important when reading that, just 7 words can be copyright infringement.
        • oh_sigh 3y ago
          It's not an argument, it is a question. I'd like to know the answer as to whether the method matters to the courts.
        • klyrs 3y ago
          I've seen many of these "AI" offerings regurgitate their training data verbatim. If chatgpt were miraculously immune to this, despite taking the same basic approach as everybody else, and without significant efforts to prevent regurgitation, then the onus is on them to prove that.
    • Dalewyn 3y ago
      Copyright laws have dealt with computer programs using bits of code or data that they had no right to incorporate. Changing the words spoken to sound sophisticated does not change anything. Obligatory IANAL.
    • Retric 3y ago
      Algorithms alone don’t qualify as transformative, which is why making an MP3 doesn’t count as a new work. Copyright is insidious because people tried all sorts of ways to get around it in the past. You can’t for example replace all the words in an article with synonyms and get a completely new work. Thus, asking ChatGPT to summarize an article is problematic because the output is based on the original work irrespective of what’s happening internally.
      • paulddraper 3y ago
        An MP3 is not a significantly different work. 1:1 substitutions of synonyms is not a significantly different work. Summaries are explicitly permitted, which is why CliffNotes, Wikipedia, etc can publish synopses of copyrighted works.
        • Retric 3y ago
          Cliff Notes books interpret, analyze, and comment on works which is allowed. Detailed summary alone doesn’t qualify for a fair use exception. Note how much of the ‘summery’ section isn’t just a listing of what happens but an interpretation of what happens: https://www.cliffsnotes.com/literature/f/fahrenheit-451/summary-and-analysis/part-1 https://www.cliffsnotes.com/literature/f/fahrenheit-451/summ... “At the same time, she also gives the reader the opportunity to see that the government has dramatically changed what its citizens perceive as their history. For example, Montag never knew that firemen used to fight actual fires...”
      • cscurmudgeon 3y ago
        > Thus, asking ChatGPT to summarize an article is problematic because the output is based on the original work irrespective of what’s happening internally. You are confusing training a model vs using it. This way a web browser can also be said to use NYT's content.
        • Retric 3y ago
          That’s in support of copyright being insidious not suggesting summaries are the only issue. The point was people seem to assume only direct copies are problematic but that’s not the case. Training/feature extraction is just as mechanical as making an MP3.
  • logicchains 3y ago
    > If you can’t use a 3rd party program even once without permission why would you assume you could use someone’s book etc to train a model without similar permission? Because you're trying build a silicon brain, and a carbon-based brain is allowed to learn from books and articles without permission (e.g. reading them at the library or online), so why shouldn't a silicon brain also be allowed to learn that way?
    • throwaway788 3y ago
      Yeah, but that carbon-based brain should have also obtained the learning material legally. Why would a “silicon-based brain” be any different.
      • cxr 3y ago
        In a discussion, one should strive not just to stick to saying things that are true, but to make sure that what they're saying is both true and relevant. People should obtain their learning material legally. True statement (at least as "true" as any opinionated value judgement can be, I guess). The question is how, though, is that statement relevant here? Even by the NYT's own telling (the way that OpenAI obtained their articles was from the NYT's website) what OpenAI did was not illegal copyright infringement. The problem is that NYT is under the impression that allowing search engines to access their paywalled stuff and hoping they don't do anything else with it besides putting up conventional, Google-style SERPs makes for copyright infringement if what one of them actually wants it for involves other reasons. It doesn't. There's an adequate legal instrument available for NYT to use if they want to enforce conditions on use: a contract. Do they have a contract that somebody violated? If not, they have no cause to go after anyone. And if they do have a contract that was violated, then that's still not copyright infringement—it's a breach of their contract.
        • Retric 3y ago
          You seem to be confused, giving someone permission to do X is only permission to do X and nothing else. If I open a store you have implied permission to enter the premises, but not stay inside for 16 hours. Further the second I ask you to leave you no longer have permission to be on the premises. The store doesn’t need to add a lock or put up signs, you’re aware you don’t have permission and that’s it.
  • salawat 3y ago
    ...Are you forgetting the fact that every clickwrapped application that doesn't have a network gated auth is just a binary patch away from being "legally impotent". In point of fact, one of what I consider the "shadow innovations" of the technical era is that we even accepted the idea of clickwrapped contracting as a valid legal construct; especially given the fact a canonical requirement of a contract "a meeting of the minds" never substantially happens. (No redline accommodation, no valid contract imo).
• shadowgovt 3y ago
  Essentially, it may very well be the case that AI resynthesis is in a category so novel that copyright (which is itself a completely artificial construct composed wholly of law, with no grounding in natural rights whatsoever) has not predicted it and is utterly mute on what should be done with it one way or the other. If that is the case, the time has come for novel law to address the issue. And that will be a hell of a political fight because it will basically boil down to who we feel should have more rights and resources in society.
• jgalt212 3y ago
  This statement ignores that fact that often, or too often, LLMs just regurgitate the training data. Direct reproduction is not a derivative work.
• tbrownaw 3y ago
  > understand the public’s emotions but recall that Is it actually the public, or is it three dinosaurs wearing a PR campaign?