I suppose we're not supposed to ask where the information in these proprietary machine learning models came from.
I suppose we're not supposed to ask where the information in these proprietary machine learning models came from.
@mhoye > IP theft
"We're all trying to find the guy who did this"
It's increasingly hard to come up with satire for this fucking industry
Let me explain.
There’s nothing to explain. You’re trying to kidnap what I’ve rightfully stolen.
@mhoye they’re giving the game away early. their desired state of things is that all of the open data sources used to train their LLMs have been scraped to death, polluted with LLM slop to the point of uselessness, or bought and made proprietary. only their degraded derivative will remain.
this is also one of the mechanisms that will be used to ensure that supposedly open source LLMs will not be a threat. all of them are derived from proprietary models, even the ones that claim not to be.
@mhoye You know, I have mixed opinions on the ethics of training LLM and image synthesis models.
I have zero mixed opinion about doing anything you can at the IO of the model to extract the trained data. Absolutely none. No compunctions whatsoever.
We can't always be expected to get along, and if we have to fight over something, it might as well be about where we have the right not to be fought. The word "mine" is simple enough to be quickly apprehended, if backed by laws enforced by the most powerful among us, their absurd hoard and inherent disregard playing a role in the everyday life when we need a fence to be respected. Reference their guns, and I don't need guns to have peace.
Hope for more would be idealistic. Let's not be silly.