Elon Musk's legal case against OpenAI
I felt like they contradicted themselves a number of times, and seem to actually support what I have read Elon saying. I don't think Elon was wrong envisioning they would need more money then they could likely raise themselves and that Tesla would be the only reasonable funding path. Of course we know now he was wrong, but he wasnt wrong statistically. How could a good businessman make a business decision assuming they could raise that much realistically?
This just sounds like good business to me. Elon appears to be acting in a fair, but capitalist manner, the same as they ended up doing themselves. Had they failed to raise sufficient funds, the project would have died, and Elon would have been right all along. It turns out they succeeded, and is the only reason they could make such a response.
I don't think anyone will be a definitive winner coming out of it. Despite Elon or OpenAI wins, their goals as per some engineers from Twitter is that AGI will be achived. And the chances are it will not be open sourced if Ilya is being OK with not being open source of it's backend tech. Ilya seems to have huge influence on the direction of where the AGI is heading in OpenAI. I haven't heard any other engineer/research enough who has that much influence there. And the speculations about Q* model being AGI achieved last year would much probably are true too. The latest updates from Anthropic Cloude 3 are also very interesting. Some researchers are claiming it is AGI level, and ahead of GPT-4 model.
I guess his case is just that try. Get as much or little information he can get through this legal battle. And try to slow down a little bit of OpenAI's efforts to make use of closed source AI which isn't actually thinking of humankind overall but profit-making machine. Elon says an ultimate-truth seeking AI is the best AGI outcome. Most probably right. While the race of these billionaire AI companies seems the looking like OpenAI is the clear winner, in the long run I still expect any and all Open Source AI models will possibly give head on competition to the closed source one. Looking at what has open source achieved in last few decades is the evidence of it. So, in that sense there is good chance Meta's Llama would be in good position, since they are already leading there. Apple has their open source framework competing against Nvidia's open source framework. They are also gonna compete hard against Nvidia. Nvidia itself is making great AI models, just not fully open source as I know so far. But are influencing other AI companies with their research, like Adobe, Stability etc. Alibaba from China is also one which is announcing big advancements almost every week. My understanding is the existing framework will be upgraded to new ones with more memory efficient and energy efficient models, that will make even the small guys at home like myself to add training data efficiently and good quality to these AI models, so in that sense any and all open source models will be gaining great advancements.
I can say that because I have seen it and experienced it already. The Stability AI released Stable Diffusion Model 1.5, then 2.0, 2.1, XL, Lightning so far. The best of their models are still based on SD 1.5! because people took their time to training their own data, merge their trained models with other people's models, made better versions of mixed models. And the evolution today we can see of these models is even better than what Midjourney and some other closed source models are generating. No need to pay any subscription fees, just a fancy Nvidia or even AMD graphics card with 16GB to 24GB VRAM and you can do all that you can expect from online models. It is not limited to image, but to audio, and vision models too. The interface for these models are continuously getting better almost every month. So I have big hopes from open source than closed source to be honest.
