It seems like the lawsuit revolves around a claim that GPT-4 is "AGI." Seems kind of dubious but, of course, when these questions get to courts who knows what will happen.
Discovery is the play here for both GPT4 and Q. It's a win/win for Elon Musk as he will either get money or get the knowledge how it's done/going to be done. I hold an opinion that GPT4 is simply an ensemble of GPT-3's with a bunch of filtering, deployment (for calculating things), and feedback mechanisms with a shitty UI tacked onto it. Q is probably that ensemble plugged into Sora somehow, to help tweak the wisdom understanding of a certain class of problems. That's why they need the GPUs so badly. And, we just saw the paper on quantization of models come out, so it's good timing to bring this claim to bear.
Elon Musk would do well to consider taking Tesla's ability to build hardware and apply it to building ASICs, because without the hardware, no amount of software discovery will net you AGI.
Musk claims the deal between OpenAI and MS is - MS gets access only to OpenAI pre-AGI tech. And he claims MS influences OpenAI board to not classify their AGI tech as AGI.
Based on that it stands to reason Musk would make a case of determining whether openai achieved AGI internally via gpt4 or q* through discovery. Maybe he can get depositions from ousted openai members to support this?
I'm not a lawyer, just trying to follow the breadcrumbs..
