Elon Musk isn’t wrong about OpenAI. It started as one thing and then became another, though that isn’t necessarily a bad thing.
The enfant terrible of the tech and social media world launched a lawsuit (PDF) late Thursday against OpenAI, and cofounders Sam Altman and Gregory Brockman, claiming that by launching a for-profit arm, OpenAI breached the terms of its original foundation agreement, which promised OpenAI would work to ensure that AI and even General AI would be developed for the betterment of humanity.
Musk, Altman, and Brockman launched OpenAI back in 2015 with that agreement in place, but after pouring millions ($44M by Musk’s count) into the non-profit and multiple disagreements about the direction of the company, Musk and OpenAI eventually parted ways.
In the meantime, OpenAI launched a profit arm, which could do a better job at funding its research than collecting donations, and it partnered up with Microsoft, which handed OpenAI a huge multi-billion investment and gained exclusive access to some of the company’s work product. That’s given Microsoft a leg up in the AI race and helped it launch Copilot (formerly Bing Chat).
In the lawsuit, which seeks unspecified damages and OpenAI’s return to its non-profit, open-source roots, Musk claims that OpenAI “has been transformed into a closed-source de facto subsidiary of the largest technology company in the world: Microsoft.”
It points to the development of GPT-4, which Musk claims can out-reason humans, and how its trajectory has not matched that of GPT-3.5. Unlike previous OpenAI GPT models, GPT-4 is not yet open-sourced. Musk claims, though, that Microsoft has access to that internal design.
Too powerful
And then there’s the generative AI model so powerful that it may have led to the temporary ouster as CEO of Sam Altman (before being returned four days later). Dubbed Q* (pronounced Q Star), Q* might be the dangerously powerful superintelligence that had OpenAI’s Chief Scientist Ilya Sutskever so worried that he sided with the board when they decided to remove Altman.
Most of that board is now gone, and Altman is once again firmly in control. Musk’s lawsuit contends that this is a recipe for disaster, claiming that the new board lacks “substantial AI expertise and, on information and belief, are ill-equipped by design to make an independent determination of whether and when OpenAI has attained AGI…”
A key element of the lawsuit is Musk’s belief that any AGI (superintelligence) development is outside the scope of the Microsoft agreement and, if OpenAI is giving the tech giant access to that technology, it’s breaking the rules.
The over-arching suit goal here is “to compel OpenAI to adhere to the Founding Agreement and return to its mission to develop AGI for the benefit of humanity, not to personally benefit the individual Defendants and the largest technology company in the world.”
To be fair, Musk is not asking for his millions back, though the level of compensation is noted to be well above “the Court’s jurisdictional minimum of $35,000”. As I see it though, Musk is just stamping his foot, arguing that OpenAI is no longer playing fair.
You see, it’s not like Musk never uses AI for commercial or for-profit purposes. His laughable Grok chatbot, the first product out of his newly-formed xAI artificial intelligence company, charges for use, or, more accurately, it asks you to pay an X (formerly Twitter) subscription fee to get its snarky answers to life’s questions.
Musk’s Tesla uses Dojo, a powerful and expensive supercomputer, to develop computer vision video processing. Dojo AI is used to develop autonomous driving systems for Teska’s equally expensive EVs.
There is some merit to Musk and others’ concerns about the inherent danger of AI run amok, but even he has admitted that the world is caught up in an AI arms race.
Have to be in it to win it
Back in 2017 when Russia President Vladimir Putin warned “Putin says the nation that leads in AI ‘will be the ruler of the world’ Musk tweeted “China, Russia, soon all countries w strong computer science. Competition for AI superiority at national level most likely cause of WW3 IMO.”
It was a warning about the fight for AI dominance, but also a signal that the war was already on. No country can afford to fall behind and, like it or not, OpenAI may give us our best chance to stay ahead.
Musk is right, though, OpenAI is trying to have it both ways: retain the altruistic sheen of a non-profit with one hand, while still collecting the receipts with the other.
Perhaps it would make more sense for OpenAI to drop the ruse and simply become what it mostly is, anyway: a for-profit company.
What we’re quickly learning is that companies that develop AI cannot be the ones who are also policing it.
Imperfect nonprofits like the Partnership for AI (which, yes, has Facebook, Amazon, Google, and IBM are among its founders) at least do not try to build the next generative platform. Instead, they focus on best practices and guidance.
In the end, Elon Musk is not a champion of AI rights or a protector of the realm. He’s a frustrated former partner who wonders why his own efforts aren’t as impactful and if he can use the courts to make himself feel better while undermining an ex-flame.