Previous work has already shown that image generators can be forced to generate examples from their training data—including copyrighted works—and an early OpenAI LLM produced contact information belonging to a researcher
You don’t seem to be able to read the articles, yet have responded with junk anyway.
You know that’s not the LLM’s ‘source’ right? It’s still output. Do you mean the training data? Is that what you mean by CoPilot should be open source? If CoPilot has learned from something GPL then everything else it outputs, or perhaps specifically its training data - should be GPL?
How exactly do you expect to see the “source” of a language model?
…
Hey does anyone want to buy a t-shirt from me with this guy’s worst comments printed on it?
I want a t-shirt with Dalton from Roadhouse on it that says “Keep it Swayze!”.
Yeah that’s not the source, that’s still output. You don’t seem to understand how LLMs work and yet have taken a bizarre stance on it anyway.
You don’t seem to be able to read the articles, yet have responded with junk anyway.
Anti Commercial-AI license
You know that’s not the LLM’s ‘source’ right? It’s still output. Do you mean the training data? Is that what you mean by CoPilot should be open source? If CoPilot has learned from something GPL then everything else it outputs, or perhaps specifically its training data - should be GPL?