cross-posted from: https://midwest.social/post/9303135
Huh, though the #ElonMusk clock is broken, this is one of the times of the day it’s still correct:
Elon Musk accused Sam Altman and OpenAI of pursuing profit over bettering humanity in a new breach of contract lawsuit filed in San Francisco Superior Court yesterday, Feb. 29.
Musk helped Altman found OpenAI as a non-profit in 2015 (Musk left the board of directors in 2018 and no longer has a stake). Central to the lawsuit is OpenAI’s “founding agreement,” which, per the lawsuit, stated the lab would build artificial general intelligence (AGI) “for the benefit of humanity,” not to “maximize shareholder profits,” and that the technology would be “open-source” and not kept “secret for propriety commercial reasons.”
Musk’s new lawsuit alleges that OpenAI has reversed course on this agreement, particularly through its $13 billion partnership with Microsoft. It further calls out the secrecy shrouding the tech behind OpenAI’s flagship Chat GPT-4 language model and major changes to the company’s board following Altman’s tumultuous hiring and re-firing last year.
“These events of 2023 constitute flagrant breaches of the Founding Agreement, which Defendants have essentially turned on its head,” the suit reads. “To this day, OpenAI, Inc.’s website continues profess that its charter is to ensure that AGI ‘benefits all of humanity.’ In reality, however, OpenAI, Inc. has been transformed into a closed-source de facto subsidiary of the largest technology company in the world: Microsoft.”
. . .
wow, that could actually be big if openAI loses! Imagine open source chatGPT and Dall-E…
I bet that you’d get the model but no data set and so it would be useless. Of course, you can feed it your own corpora if you like.
it wouldnt be useless, most “open source” models work like this. But yeah, having access to the dataset is very important, and in my opinion open source licenses requires its inclusion. However, no legal battles have been fought over that fact yet…
I guess I’m saying that it would be useless unless you are willing to train it, assuming that the training corpus isn’t included.
I think it will be released pre-trained if they are forced to make it open source, otherwise they are not actually releasing chatgpt/dall-e, just the underlying technology they used to make it. No idea however if musks lawyers know that.