Litigation between the world's richest man and one of the fastest-growing companies in history is automatically fascinating in its own right. But while the claims are still to be proved, the case has already shed an interesting batch of emails from Elon Musk, Sam Altman, and others during the early days of OpenAI. Here are some of the more interesting snippets we unearthed while browsing through their correspondence.
Remember that these emails were unearthed in the process of an effort to show that OpenAI somehow was violating antitrust law (a frankly implausible charge). Also, they reveal in part how Musk felt when OpenAI abandoned its original vision of being a nonprofit with the Tesla CEO as its leader.
They do not tell the whole story but are interesting in their own right.
Perhaps the most interesting single email is from former chief scientist Ilya Sutskever as he explains the team's qualms over Musk's leadership of the company.
The current structure gives you a pathway whereby you end up with unilateral absolute control over the AGI [artificial general intelligence]. You stated that you don't want to control the final AGI, but during this negotiation, you've shown to us that absolute control is extremely important to you.
For example, you said you have to be the CEO of the new company so everyone will know that you are the person in charge, and yet you also stated that you hate being a CEO and would much rather not be.
So, we are going to be concerned that as the company finally starts to make the real moves toward true AGI, you will want to retain your absolute control over the company in its current state against any intentions to the contrary.
The purpose of OpenAI is to make the future good and to avoid an AGI dictatorship. You are worried that Demis [Hassabis, at Google-owned DeepMind] could create an AGI dictatorship. So do we. So it is a bad idea to create a structure where you could become a dictator if you chose to, especially given that we can create some other structure that avoids this possibility.
This isn't all about corporate control; Sutskever fears that the existential AI threat being created has only one person in its way.
Sutskever also states his concerns with regards to Altman, using words largely the same that the board would use when casting its judgment on him as "not consistently candid":
We haven't been able to fully trust your judgements throughout this process, because we don't understand your cost function.
We don't understand why the CEO title matters to you, anyway. Your reasons of choice have changed, and it's hard to really get at what it is driving.
Is AGI really your bottom-line motivation? How does that relate to your political goals?
Considering how things ended and the turns Altman took for the company toward a far more traditional enterprise SaaS model, we feel like his aim was more about business than philosophy.
One little-known fact is that, as of 2017, OpenAI was seriously considering buying chipmaker Cerebras or somehow merging with it, perhaps using Tesla's resources somehow. As Sutskever writes in an email:
In the event we decide to buy Cerebras, my strong sense is that it'll be done through Tesla .
They did not take this action, although the reason why is not in these emails.
This, at any rate, was when Musk was wooing to make OpenAI one of his many investment assets, and the founders were amenable to that. As OpenAI co-founder Andrej Karpathy wrote :
The best I can imagine, as mentioned earlier, would be for OpenAI to attach to Tesla as its cash cow. (.) If we execute this very well, the transportation industry is large enough that we could grow the market cap of Tesla to high O(~100K), and then use those revenues to fund the AI work at the appropriate scale.
Again, that didn't happen for a variety of reasons that, in hindsight, seem fairly obvious. Tesla's market cap did, in fact, rise, but the self-driving side of things - which Karpathy sought to speed up later in his career when he began working at Tesla - has been harder than anyone had anticipated, and hasn't yet added meaningfully to Tesla's revenue.
As for monetization, Microsoft was involved at least as far back as 2016, committing to provide OpenAI with $60 million in compute on Azure in return for, among other things, that the companies "evangelize" one another. Nobody seemed to care about this kind of corporate back-scratching, and Musk said it made him "nauseous."
They ultimately paid much more but with no obligation on either side. "Would be worth way more than $50M not to seem like Microsoft's marketing bitch," wrote Musk.
Finally, a small crumb dropped by board member Shivon Zilis (who would eventually be the mother of three of Elon Musk's children): Valve founder Gabe Newell was, apart from being an early donator to the project, on Altman and Greg Brockman's "informal advisory board." I'm not sure what role he had or plays there. I reached out to Newell for comment.