Comment on OpenAI says it’s “impossible” to create useful AI models without copyrighted material

Powderhorn@beehaw.org ⁨11⁩ ⁨months⁩ ago

Any reasonable person can reach the conclusion that something is wrong here.

What I’m not seeing a lot of acknowledgement of is who really gets hurt by copyright infringement under the current U.S. scheme. (The quote is obviously directed toward the UK, but I’m reasonably certain a similar situation exists there.)

Hint: It’s rarely the creators, who usually get paid once while their work continues to make money for others.

Let’s say the New York Times wins its lawsuit. Do you really think the reporters who wrote the infringed-upon material will be getting royalty checks to be made whole?

This is not OpenAI vs creatives. OK, on a basic level it is, but expecting no one to scrape blogs and forum posts rather goes against the idea of the open internet in the first place. We’ve all learned by now that what goes on the internet stays there, with attribution totally optional unless you have a legal department. What’s novel here is the scale of scraping, but I see some merit to the “transformational” fair-use defense given that the ingested content is not being reposted verbatim.

This is corporations vs corporations. Framing it as millions of people missing out on what they’d have otherwise rightfully gotten is disingenuous.

source
Sort:hotnewtop