[ad_1]
A lawsuit filed by the world’s richest man against one of the fastest-growing companies of all time is necessarily interesting stuff. But while these claims have yet to be proven, the case has already uncovered a set of emails between Elon Musk, Sam Altman, and others during the early days of OpenAI. Below are some of the more interesting excerpts we found while following their correspondence.
Keep in mind that these emails were revealed as part of an attempt to prove that OpenAI is somehow violating antitrust law (which is, frankly, an implausible claim). Musk also reveals somewhat that he felt betrayed when OpenAI abandoned its original vision of being a non-profit with the Tesla CEO as its leader.
They don’t tell the whole story, but they’re still interesting in their own right.
Perhaps the most interesting email is from former chief scientist Ilya Sutskever explaining the team’s concerns about Musk as the company’s leader:
The current structure provides you with a path to end up with absolute unilateral control of AGI (Artificial General Intelligence). You have stated that you do not want to control the final AGI, but during these negotiations, you have shown us that absolute control is very important to you.
For example, you said that you need to become CEO of the new company so everyone knows you are the person in charge, even though you also mentioned that you hate the CEO position and would rather not be CEO.
Thus, we are concerned that while the Company makes real progress towards AGI, it will choose to retain ultimate control of the Company despite current intent to the contrary.
The goal of OpenAI is to make the future good and avoid the dictatorship of general artificial intelligence. You worry that Demis (Hasabis, at Google-owned DeepMind) could create an artificial general intelligence dictatorship. And so are we. So, it’s a bad idea to create a structure where you can become a dictator if you so choose, especially since we can create another structure that avoids that possibility.
This is not entirely about corporate control; Sutskever is concerned about an existential AI threat being created with only one person in the way.
Sutskever also expressed concerns about Altman, using words very similar to what the board would later use while accusing him of not being “consistently honest”:
We have not been able to fully trust your judgment throughout this process, because we do not understand your cost function.
We don’t understand why the title of CEO is so important to you. The reasons I mentioned have changed, and it’s hard to really understand why.
Is Artificial General Intelligence (AGI) really your primary motivation? How does it relate to your political goals?
Given the way things turned out and Altman steered the company toward a more traditional enterprise SaaS site, it seems his goal was more commercial than philosophical.
One interesting bit of news is that as early as 2017, OpenAI was seriously considering buying chipmaker Cerebras, or merging with it in some way, and perhaps using Tesla’s resources in some way. As Sutskever says:
If we decide to buy Cerebras, my strong feeling is that it will be done through Tesla.
They ended up not going through with it, although the reason is not in these emails.
Incidentally, this happened when Musk was seeking to make OpenAI just one of his many properties, and leaders were open to the possibility. As OpenAI co-founder Andrei Karpathy wrote:
The most promising option I can think of, as I mentioned earlier, is for OpenAI to tie up with Tesla as its cash cow. (…) If we do it really well, the transportation industry is big enough that we can grow Tesla’s market cap to high O (~100K), and use that revenue to fund AI work at the right scale.
Again, this did not happen for many reasons that seem obvious in hindsight. In fact, Tesla’s market capitalization increased, but the self-driving side of things — which Karpathy later aimed to accelerate when he took a job at Tesla — proved more difficult than expected, and has yet to contribute meaningfully to Tesla’s revenue.
In terms of making money, Microsoft has been in the mix since as early as 2016, offering OpenAI $60 million worth of compute on Azure in exchange for, among other things, the companies “evangelizing” to each other. No one seemed interested in this kind of corporate criticism, and Musk wrote that it made him “sick.”
They eventually ended up paying a much larger amount but with no commitment on either side. “It would be worth more than $50 million to not look like Microsoft’s marketing whore,” Musk wrote.
Finally, board member Shivon Zillis (who later became the mother of three of Musk’s children) mentioned a little something: Valve founder Gabe Newell, in addition to being a donor to the project in the early days, was at the expense of Altman and Greg Brockman. “Informal advisory board.” It is not clear what role he or she had in daily life there. I asked Newell for comment.
[ad_2]