4 ms·

What’s most frustrating to me is the ridiculous false equivalency people make between computers and humans “Babies learn from their environment why not GPT?”

What’s most frustrating to me is the ridiculous false equivalency people make between computers and humans “Babies learn from their environment why not GPT?” “Human teachers remix stuff for their lessons all the time, why can’t GPT?” Computers are not humans. Computers are not humans. Computers are not humans. Good grief.

empath-nirvana 3y ago
OpenAI is a corporation, which means it's a person under US law. So yes, it's relevant as to whether a person can remix copyrighted material or learn from copyrighted material, because the same laws that apply to people apply to corporations. Moreover, the AI the product of the collective action of _people_, it wasn't formed out of nothing.
- probably_wrong 3y ago
  > OpenAI is a corporation, which means it's a person under US law. No, corporations are not people. They have some of the rights a person has, but not all. For example, from [1], "it is firmly established that a corporation has no fifth amendment protection against self-incrimination". [1] https://law.justia.com/cases/federal/appellate-courts/F2/515/295/292973/ https://law.justia.com/cases/federal/appellate-courts/F2/515...
  - empath-nirvana 3y ago
    Corporate rights are basically limited to the rights that any collective group of people would have, so self-incrimination doesn't make sense, but it doesn't get around the broader point that the same corporate personhood that allows corporations to be sued for copyright infringement, allows them to use the same defenses that groups of individuals (ie, co-authors) could use for copyright infringement. Could you sue "Alice and Bob" for copyright infringement because they watched a lot of movies before writing their screenplay? Nobody writes a screenplay from first principles. They're all based on all the other movies they've seen to some extent.
- zilti 3y ago
  BS doesn't become true through repetition - unless you're an AI, that is.
skepticATX 3y ago
Current AI discourse has made me realize that there is quite the dichotomy between people who see humans as nothing more than biological machines that will soon be obsoleted by synthetic machines, and people who see humans (as Kant would say) as an end. I think that it's sad that the leading AI labs seem to mostly be run by people in the former camp. It doesn't have to be this way.
- elzbardico 3y ago
  they're only in it for the money.
- logicchains 3y ago
  > people who see humans (as Kant would say) as an end Kant's arguments would apply just as much to intelligent, self conscious machines as to humans; nowhere does he base his conclusions on something as dumb as "because we're biological".
  - skepticATX 3y ago
    Of course. And if machines become conscious (I don’t think we should ever build something like this even if we can), I’d be the first to advocate that we give them moral consideration. Right now though, we should put humans, and animals for that matter, before anything else.
    - 93po 3y ago
      This feels overly simplistic to me because there's no way to define "conscious"
mdorazio 3y ago
> Computers are not humans. And? Why does that invalidate either of your two examples? You need to answer "why not GPT?" with something more substantial than "because it's not a human."
- xbar 3y ago
  Oh. This one is easy. ChatGPT a commercial software product that generates wealth for its owners, so allowing commercial products to steal and promote others' content on their own is an agreed-upon violation of rights. Let's look at an example: Google doesn't get to index and show New York Times content directly in its search results. If you think of ChatGPT more like Google maybe that will help you keep it straight.
  - hhjinks 3y ago
    That wouldn't be legal even if GPT was a person. It being software doesn't play into your argument at all.
  - 93po 3y ago
    I'm a biological cog that generates wealth for my owners (i mean employers) and I use others' copyrighted content to learn how to do so
  - mdorazio 3y ago
    Ok. So GPT is a tool and therefore the legality is entirely tied to the use of its outputs. And we already have a large body of law around that for IP. I’m still not understanding your argument. And Re: Google. I agree. Which is why the case law around Google books showing full excerpts is important to look at.
elzbardico 3y ago
People opinions don't come out of a vacuum and they usually do a lot of rationalization when it comes to defending their own prejudices and whatever matches their self interest. This is HN, lots of people here see LLMs the same way miners saw the gold rush. When you deal with people dreams of richness, you can be pretty sure they would move us all right into a skynet situation, rationalizing every single fucking step towards our demise.
logicchains 3y ago
>Computers are not humans. A computer that can reason and is self aware like a human is _morally equivalent_ to a human unless your morality is based on some kind of human supremist philosophy. And I imagine in future human supremacists are going to be regarded the same way white supremacists are regarded currently.
- golly_ned 3y ago
  This is also a speciesist position for making human-like reason and self-awareness the selector for moral status, which selects out non-human animals.
xpe 3y ago
Yes, the lack of clarity between human characteristics and machine characteristics here is problematic. For example, there are two particularly salient differences at hand: (1) the ability to make cheap digital copies of machine intelligences; (2) the rights granted to humans. This lack of clarity is emblematic of how legal mechanisms operate. Loosely speaking, in comparison with computational and philosophical approaches, the law is quite lazy at drawing distinctions. Disagreements get deferred using a pre-determined legal process. One might say the lazy-approach of the law is "by design". I'm not sure; I'd need to dig into the history a lot more to assess.
tensor 3y ago
What frustrates me the most are comments like yours that contribute nothing whatsoever to the debate about the limits of copyright. Humans are not special, get over it.
duskwuff 3y ago
I've already seen one person today suggest that discontinuing an generative image model that produces problematic output is comparable to murdering an artist. It's pretty absurd.

empath-nirvana 3y ago

OpenAI is a corporation, which means it's a person under US law. So yes, it's relevant as to whether a person can remix copyrighted material or learn from copyrighted material, because the same laws that apply to people apply to corporations. Moreover, the AI the product of the collective action of _people_, it wasn't formed out of nothing.

probably_wrong 3y ago
> OpenAI is a corporation, which means it's a person under US law. No, corporations are not people. They have some of the rights a person has, but not all. For example, from [1], "it is firmly established that a corporation has no fifth amendment protection against self-incrimination". [1] https://law.justia.com/cases/federal/appellate-courts/F2/515/295/292973/ https://law.justia.com/cases/federal/appellate-courts/F2/515...
- empath-nirvana 3y ago
  Corporate rights are basically limited to the rights that any collective group of people would have, so self-incrimination doesn't make sense, but it doesn't get around the broader point that the same corporate personhood that allows corporations to be sued for copyright infringement, allows them to use the same defenses that groups of individuals (ie, co-authors) could use for copyright infringement. Could you sue "Alice and Bob" for copyright infringement because they watched a lot of movies before writing their screenplay? Nobody writes a screenplay from first principles. They're all based on all the other movies they've seen to some extent.
zilti 3y ago
BS doesn't become true through repetition - unless you're an AI, that is.

skepticATX 3y ago

Current AI discourse has made me realize that there is quite the dichotomy between people who see humans as nothing more than biological machines that will soon be obsoleted by synthetic machines, and people who see humans (as Kant would say) as an end. I think that it's sad that the leading AI labs seem to mostly be run by people in the former camp. It doesn't have to be this way.

elzbardico 3y ago
they're only in it for the money.
logicchains 3y ago
> people who see humans (as Kant would say) as an end Kant's arguments would apply just as much to intelligent, self conscious machines as to humans; nowhere does he base his conclusions on something as dumb as "because we're biological".
- skepticATX 3y ago
  Of course. And if machines become conscious (I don’t think we should ever build something like this even if we can), I’d be the first to advocate that we give them moral consideration. Right now though, we should put humans, and animals for that matter, before anything else.
  - 93po 3y ago
    This feels overly simplistic to me because there's no way to define "conscious"

mdorazio 3y ago

> Computers are not humans. And? Why does that invalidate either of your two examples? You need to answer "why not GPT?" with something more substantial than "because it's not a human."

xbar 3y ago
Oh. This one is easy. ChatGPT a commercial software product that generates wealth for its owners, so allowing commercial products to steal and promote others' content on their own is an agreed-upon violation of rights. Let's look at an example: Google doesn't get to index and show New York Times content directly in its search results. If you think of ChatGPT more like Google maybe that will help you keep it straight.
- hhjinks 3y ago
  That wouldn't be legal even if GPT was a person. It being software doesn't play into your argument at all.
- 93po 3y ago
  I'm a biological cog that generates wealth for my owners (i mean employers) and I use others' copyrighted content to learn how to do so
- mdorazio 3y ago
  Ok. So GPT is a tool and therefore the legality is entirely tied to the use of its outputs. And we already have a large body of law around that for IP. I’m still not understanding your argument. And Re: Google. I agree. Which is why the case law around Google books showing full excerpts is important to look at.

elzbardico 3y ago

People opinions don't come out of a vacuum and they usually do a lot of rationalization when it comes to defending their own prejudices and whatever matches their self interest. This is HN, lots of people here see LLMs the same way miners saw the gold rush. When you deal with people dreams of richness, you can be pretty sure they would move us all right into a skynet situation, rationalizing every single fucking step towards our demise.

logicchains 3y ago

>Computers are not humans. A computer that can reason and is self aware like a human is _morally equivalent_ to a human unless your morality is based on some kind of human supremist philosophy. And I imagine in future human supremacists are going to be regarded the same way white supremacists are regarded currently.

golly_ned 3y ago
This is also a speciesist position for making human-like reason and self-awareness the selector for moral status, which selects out non-human animals.

xpe 3y ago

Yes, the lack of clarity between human characteristics and machine characteristics here is problematic. For example, there are two particularly salient differences at hand: (1) the ability to make cheap digital copies of machine intelligences; (2) the rights granted to humans. This lack of clarity is emblematic of how legal mechanisms operate. Loosely speaking, in comparison with computational and philosophical approaches, the law is quite lazy at drawing distinctions. Disagreements get deferred using a pre-determined legal process. One might say the lazy-approach of the law is "by design". I'm not sure; I'd need to dig into the history a lot more to assess.

tensor 3y ago

What frustrates me the most are comments like yours that contribute nothing whatsoever to the debate about the limits of copyright. Humans are not special, get over it.

duskwuff 3y ago

I've already seen one person today suggest that discontinuing an generative image model that produces problematic output is comparable to murdering an artist. It's pretty absurd.