Ai Artificial Intelligence Archive

Also, Musk gave no guarantee his proposed control of the board would diminish over time. “I found it to be aggressive because I knew that Mr Musk had many other obligations in many other companies that the was running that were much larger than OpenAI,” he said. He also didn’t like the proposal that Tesla take over OpenAI. “It would be on some level, it would be like, it would kill a dream,” he said. “When one starts a company, one has dreams for a company to flourish and do different things, and in general being absorbed into another company means to give up that dream.”

We are now at “the blip” again. The board chose “not consistently candid” carefully, he says. Sutskever said he prepared a document of incidents with Altman, with some other people at OpenAI. Altman has a pattern of lying and pitting executives against each other, “this leads to tremendous loss of productivity,” trust, and difficulty creating safe AGI.

The family of a victim of April’s mass shooting at Florida State University is suing OpenAI over its chatbot’s alleged role in encouraging the attack, which is already being probed by Florida’s attorney general. OpenAI spokesperson Drew Pusateri called the shooting a “tragedy” but said “ChatGPT is not responsible for this terrible crime.” More from Pusateri:

Our next witness is Ilya Sutskever. This promises to be more interesting, I think.

He says he provides suggestions a lot. And they’re just suggestions, including in the case of OpenAI. While Microsoft put forward 14 names, none of those names were added to the new OpenAI board except Sue Desmond-Hellman. CEO of the Bill and Melinda Gates Foundation, who was added later.

He’s explaining that firing a CEO is “a fairly big things” and when he didn’t get any details on why Altman was fired, he felt that OpenAI’s board was “sort of amateur city as far as I was concerned.”

We also discover that Nadella has no idea if Altman and Musk talked about Musk’s mean Microsoft/OpenAI tweets. As a side note, we are referring exclusively to “Twitter” and “tweets” even for posts after the X rebrand.

I recognize Musk has identified this as the spot where he was convinced he’d been swindled. But the main thing I’m hearing here is that OpenAI was unstable. I don’t know what the correct way to respond to that kind of chaos is, especially if you have a partnership with the company that appears to be rapidly imploding. Trying to suggest that Nadella’s attempts to stabilize the company and preserve his investment was somehow improper seems weird? This is a genuinely odd situation, not necessarily one you run into in business school. (Well, until then anyway.)

“Pretty definitive statement as to why they fired him, right?” Well, Mr. Molo, no. I remember a great deal of speculation about what this meant; for instance, if Altman had gotten up to something criminal. One of those moments where I wonder what the jury is thinking, since I doubt they recall any of this.

Nadella is cooler than everyone else on cross, but he’s still getting worked up. Molo just got told by the judge that his question was argumentative. Molo is yelling about how the risk MSFT took on OpenAI was “prudent.” Nadella notes it’s still a risk. “At the time the was a risk it could to to zero,” Nadella says. “It was a calculated risk.” Molo, of course, is focused on projected return, of $92+ billion.

Molo is asking about this line Nadella wrote in a 2022 email, and it’s very funny. Nadella has explained a couple times that to him it’s about IP rights. “The context for me was making sure that Microsoft was benefitting from the IP rights that we had because that’s what happened in the case of Microsoft and IBM.” Molo then says, didn’t Microsoft become more prominent and important than IBM? Nadella agrees.

OpenAI is expanding its enterprise arm with the launch of the OpenAI Deployment Company, which will work with businesses to build, test, and deploy AI systems tailored to their needs. The new company received $4 billion in investments and has a $10 billion pre-money valuation, according to Axios.

Nadella’s attitude is not notably different than it was on direct. We are talking about donating commute to OpenAI from Azure services. From a 1/12/18 email: “The 2016 deal, where they agreed to pay $10M for $60M in Azure services was projected at a $15M loss over 3 years, given assumed usage profile. They’ll have consumed all the usage in ½ that time.” The funny recurring theme here through the trial is the gaping maw of compute required is always bigger than whatever huge number people had projected. Interesting to know ahead of AI IPOs!

He misstated at first that he didn’t currently serve on any such boards, though he is a trustee at the University of Chicago, which is a nonprofit.

Microsoft CEO Satya Nadella was just asked to explain what Copilot is during the Musk v. Altman trial. “Copilot is an AI assistant, similar to ChatGPT.” That’s funny because you won’t find a mention of ChatGPT in Microsoft’s Copilot marketing materials.