5 ms·

> 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 permi

> 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.
    - cxr 3y ago
      Ditch the swipes; I'm not the one confused here. We're talking about copyright infringement, not trespass. (Your analogy is bad and doesn't hold up. Copyright doesn't grant rightsholders control of the sort required here. It grants them the right to make and distribute copies. It doesn't grant the right to undistribute copies when it turns out they don't like what someone is doing with them.)
      - Retric 3y ago
        OpenAI doesn’t maintain a copy. A core argument OpenAI is making is transitory copying is allowed as long training is fair use. But a permanent copy stored in a training database isn’t transitory and would itself be copyright infringement, so they don’t do that. Thus training each version requires permission to download a new copy, which they now lack.
        cxr 3y ago
        EDIT: The comment I first responded to (only partially reconstructible from quoted parts below) has been edited to something to something almost completely different. Very uncool. > ChatGPT keeps redownloading works to avoid the issue of keeping permeant copy’s of the training material I don't know enough about how ChatGPT works to know whether or not that's true, but from an engineering standpoint it certainly sounds wrong because of how insane it would be if true, and I'm not at all convinced that you're right about this given how poorly you understand the other stuff you're trying to argue that I happen to know you're wrong about, so it doesn't seem unwise to conclude that the same is probably true about your claims here. But it doesn't matter, anyway: > which would be obvious copyright infringement Wrong. OpenAI keeps asking for copies, NYT keeps giving them out (whether/despite OpenAI having/not having "permission" or not). Not copyright infringement, let alone "obvious copyright infringement". This is going to be my last response that takes a substantial amount of effort to compose. Arguing with /r/confidentlyincorrect-tier zero-effort Gish gallops* is not a good use of my time. * especially from someone shameless about editing their posts after the fact to make them diverge substantially from what was originally written
the_snooze 3y ago
A change in scale is a change in kind. There's a real qualitative difference between a party popper and a flashbang even though they're both "just" generators of light and sound. It's the same with regular ol' human learning---for which we have millenia of understanding and norms---and comparatively low-cost global-scale machine learning. The scale of the latter obliterates the structures we've built under the assumption of the former, so you can't just claim they're the same.
itishappy 3y ago
Our laws for carbon-based brains were designed pretty explicitly for carbon-based brains. Why assume these completely different things should be (legally) equivalent despite only superficial similarities?
- tanseydavid 3y ago
  To me, the similarities do NOT seem superficial in any way.
  - itishappy 3y ago
    I'd argue you're not looking very deep. On a physical level, chips and brains are entirely dissimilar. Some algorithms may share the same names, but that's about it. Brains aren't doing matrix multiplications all the time and chips don't use the Krebs cycle.
- nickthegreek 3y ago
  In America, companies are people, no brain required and are protected in many ways because of that classification.

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.
  - cxr 3y ago
    Ditch the swipes; I'm not the one confused here. We're talking about copyright infringement, not trespass. (Your analogy is bad and doesn't hold up. Copyright doesn't grant rightsholders control of the sort required here. It grants them the right to make and distribute copies. It doesn't grant the right to undistribute copies when it turns out they don't like what someone is doing with them.)
    - Retric 3y ago
      OpenAI doesn’t maintain a copy. A core argument OpenAI is making is transitory copying is allowed as long training is fair use. But a permanent copy stored in a training database isn’t transitory and would itself be copyright infringement, so they don’t do that. Thus training each version requires permission to download a new copy, which they now lack.
      - cxr 3y ago
        EDIT: The comment I first responded to (only partially reconstructible from quoted parts below) has been edited to something to something almost completely different. Very uncool. > ChatGPT keeps redownloading works to avoid the issue of keeping permeant copy’s of the training material I don't know enough about how ChatGPT works to know whether or not that's true, but from an engineering standpoint it certainly sounds wrong because of how insane it would be if true, and I'm not at all convinced that you're right about this given how poorly you understand the other stuff you're trying to argue that I happen to know you're wrong about, so it doesn't seem unwise to conclude that the same is probably true about your claims here. But it doesn't matter, anyway: > which would be obvious copyright infringement Wrong. OpenAI keeps asking for copies, NYT keeps giving them out (whether/despite OpenAI having/not having "permission" or not). Not copyright infringement, let alone "obvious copyright infringement". This is going to be my last response that takes a substantial amount of effort to compose. Arguing with /r/confidentlyincorrect-tier zero-effort Gish gallops* is not a good use of my time. * especially from someone shameless about editing their posts after the fact to make them diverge substantially from what was originally written

the_snooze 3y ago

A change in scale is a change in kind. There's a real qualitative difference between a party popper and a flashbang even though they're both "just" generators of light and sound. It's the same with regular ol' human learning---for which we have millenia of understanding and norms---and comparatively low-cost global-scale machine learning. The scale of the latter obliterates the structures we've built under the assumption of the former, so you can't just claim they're the same.

itishappy 3y ago

Our laws for carbon-based brains were designed pretty explicitly for carbon-based brains. Why assume these completely different things should be (legally) equivalent despite only superficial similarities?

tanseydavid 3y ago
To me, the similarities do NOT seem superficial in any way.
- itishappy 3y ago
  I'd argue you're not looking very deep. On a physical level, chips and brains are entirely dissimilar. Some algorithms may share the same names, but that's about it. Brains aren't doing matrix multiplications all the time and chips don't use the Krebs cycle.
nickthegreek 3y ago
In America, companies are people, no brain required and are protected in many ways because of that classification.