3 ms·

You're not even sure this is a copyright issue in the US yet, let alone every other country on the planet. So yes, if you don't want it freely consumed, don't f

by flir 3y ago

You're not even sure this is a copyright issue in the US yet, let alone every other country on the planet. So yes, if you don't want it freely consumed, don't freely publish it.

spencerflem 3y ago
Wouldn't that be worse? Then instead of there being some way to get the information (pay them) there's now no way to get it.
- flir 3y ago
  I'm in "I just want to watch the world burn" mode over this right now, to be honest. If all that's left on the open internet is SEO spam, and all the chatbots can train on is SEO spam, then so be it. We didn't deserve to have nice things.
  - spencerflem 3y ago
    Yea the world sucks, tech companies are making it worse, the economy is built to funnel money to whoever can suck the most value out of the commons for themselves, climate change will ruin everything, hug your favorite endangered species while you still can, and the internet is best at spreading lies I don't see how arguing that it should be harder for journalists to get paid helps anything though
    - flir 3y ago
      Accelerationism, obvs. (Except I'm pretty sure that doesn't work either).
CharlesW 3y ago
> You're not even sure this is a copyright issue in the US yet… We are sure that OpenAI is building its business on copyrighted content, under the defense that to do otherwise is "impossible." The courts have not decided whether this is "fair use". However, the four factors make it pretty clear.
- flir 3y ago
  Before that you'd have to establish that whatever the chatbots are doing is "use" in the sense that "fair use" means it. The act of reading a book or looking at a painting isn't protected by fair use. If it is a use that falls under fair use, I'm thinking about the Google Books case, where an act that is far less transformative (digital archiving and search) was found to be fair use (https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,_Inc https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,....) (It's going to be "who's got more money?", isn't it? Fun fun fun.)
  - CharlesW 3y ago
    > The act of reading a book or looking at a painting isn't protected by fair use. This is a very strange way to look at it. If I encode Oppenheimer to MPEG-4, do I now own that IP? In both cases, I'm looking at the source and encoding it to a compressed representation. A couple notes on "transformative": (1) It's just one of the four factors. (2) It means that you added new expression or meaning to it. By definition, an LLM can't create anything it wasn't trained on — it's pattern recognition, interpolation, and recombination.
    - fragmede 3y ago
      When an LLM writes an essay about, say, Oppenheimer, the essay it spits out isn't, word for word, in the training data. it's created this new essay. Following that, asking it to write an essay about Doofenhimer, a fictional contemporary of Oppenheimer that I just invented, where was it trained on for that? to say it can't create anything when, given the depth of the training data, it seems capable of creating things, seems like an underestimation of its abilities, and moreover, what it actually means to be creative. when I'm being creative, there are connections being made between completely unrelated things that I've been exposed to at some point in my past which I combine in creative ways to create new things. if it's all just pattern recognition, interpolation, and recombination, then either nothing new can ever be created by humans doing exactly that, or LLMs are also creating new things. Are Salvador Dali's paintings not creative because the idea of clocks and the concept of melting aren't new? https://chat.openai.com/share/2b9a8245-2063-4722-81e8-20d10ce26104 https://chat.openai.com/share/2b9a8245-2063-4722-81e8-20d10c...

by flir 3y ago

You're not even sure this is a copyright issue in the US yet, let alone every other country on the planet. So yes, if you don't want it freely consumed, don't freely publish it.

spencerflem 3y ago

Wouldn't that be worse? Then instead of there being some way to get the information (pay them) there's now no way to get it.

flir 3y ago
I'm in "I just want to watch the world burn" mode over this right now, to be honest. If all that's left on the open internet is SEO spam, and all the chatbots can train on is SEO spam, then so be it. We didn't deserve to have nice things.
- spencerflem 3y ago
  Yea the world sucks, tech companies are making it worse, the economy is built to funnel money to whoever can suck the most value out of the commons for themselves, climate change will ruin everything, hug your favorite endangered species while you still can, and the internet is best at spreading lies I don't see how arguing that it should be harder for journalists to get paid helps anything though
  - flir 3y ago
    Accelerationism, obvs. (Except I'm pretty sure that doesn't work either).

CharlesW 3y ago

> You're not even sure this is a copyright issue in the US yet… We are sure that OpenAI is building its business on copyrighted content, under the defense that to do otherwise is "impossible." The courts have not decided whether this is "fair use". However, the four factors make it pretty clear.

flir 3y ago
Before that you'd have to establish that whatever the chatbots are doing is "use" in the sense that "fair use" means it. The act of reading a book or looking at a painting isn't protected by fair use. If it is a use that falls under fair use, I'm thinking about the Google Books case, where an act that is far less transformative (digital archiving and search) was found to be fair use (https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,_Inc https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,....) (It's going to be "who's got more money?", isn't it? Fun fun fun.)
- CharlesW 3y ago
  > The act of reading a book or looking at a painting isn't protected by fair use. This is a very strange way to look at it. If I encode Oppenheimer to MPEG-4, do I now own that IP? In both cases, I'm looking at the source and encoding it to a compressed representation. A couple notes on "transformative": (1) It's just one of the four factors. (2) It means that you added new expression or meaning to it. By definition, an LLM can't create anything it wasn't trained on — it's pattern recognition, interpolation, and recombination.
  - fragmede 3y ago
    When an LLM writes an essay about, say, Oppenheimer, the essay it spits out isn't, word for word, in the training data. it's created this new essay. Following that, asking it to write an essay about Doofenhimer, a fictional contemporary of Oppenheimer that I just invented, where was it trained on for that? to say it can't create anything when, given the depth of the training data, it seems capable of creating things, seems like an underestimation of its abilities, and moreover, what it actually means to be creative. when I'm being creative, there are connections being made between completely unrelated things that I've been exposed to at some point in my past which I combine in creative ways to create new things. if it's all just pattern recognition, interpolation, and recombination, then either nothing new can ever be created by humans doing exactly that, or LLMs are also creating new things. Are Salvador Dali's paintings not creative because the idea of clocks and the concept of melting aren't new? https://chat.openai.com/share/2b9a8245-2063-4722-81e8-20d10ce26104 https://chat.openai.com/share/2b9a8245-2063-4722-81e8-20d10c...