Gosh, sounds like fair use to me. It's not like OpenAI / Microsoft cared one bit about licenses or copyright when they absorbed my entire legal blog into their LLMs via Common Crawl. Why should their probabilistic LLM outputs get better copyright protection than my original human work?
Comments
It is delicious!
It's like IP only matters if its record companies or publishers not artists. Data only matters if it's big tech, not people.
The purpose of IP is to reward people (living humans) for their work, so that they don't put time & effort into a project only to have it immediately stolen by someone else.
To secure a living & prevent freeloading.
Giving their output any sort of IP protection would be akin to giving them everything not currently protected or already in the public domain.
Instant output, instant public domain.
Doesn't intellectual property protection require, as a critical element, that that work seeking protection is novel?
And isn't LLM AI product, in fact, imitative and not authentically novel?
No human artist is going to give their work a leg coming out of its armpit.
I mean, it doesn’t help, certainly, but it’s a baked in feature of how the tech works. It’ll never get better without crude kludges to fake it.
It takes a thousand different images, overlays them on a centerpoint, blurs the lines and calls it "novel".
The result of that is that fingers aren't easy to overlay, so you get hands with 6 fingers.
And you're correct, I am. And that is ... amazing frankly.
high, HIGH irony if so.
China bad when they do it. Cause it’s a communism when they do it.
;)
Not that it matters. They didn't care when Mistral did this, but now that a big bad Chinese company is doing it, it's newsworthy... #wtf
They're just made that Deepseek did it faster and cheaper.
MURKA
“Yeehaw this is great! You can’t stop progress or capitalism, plebes!! Woohoo!”
Microsoft when Deepseek steals their work:
“This is the greatest affront to intellectual property rights in the history of the world! How dare they! HOW! DARE! THEY!”