(Bloomberg) -- Two days after Sam Altman reached an agreement with OpenAI to return as its chief executive, he spent part of his Thanksgiving with Adam D’Angelo, one of the company’s board members who had fired him the week prior.Their hours-long meeting, which Altman called “really nice,” highlights D’Angelo’s unique role in a corporate drama that has captivated Silicon Valley — and the importance of their relationship in restoring some stability at the world’s best-known artificial intelligence startup. As part of the deal to bring Altman back, the board is set to be completely overhauled, with one exception: D’Angelo will stay on as a director.

D’Angelo’s staying power at the company may have surprised some given his part in ousting Altman for not being “consistently candid in his communications with the board.” In an interview during the leadership drama, Vinod Khosla, founder of Khosla Ventures — one of OpenAI’s earliest investors — said he believed D’Angelo remained firm on his decision, despite how much the move riled investors and employees. After nearly all of OpenAI’s staff threatened to quit, however, D’Angelo became a key figure in negotiations with Altman about his return.

D’Angelo’s involvement throughout the OpenAI saga has brought new attention and scrutiny to a longtime Silicon Valley insider. As the co-founder of Quora, a question-and-answer website, and an early Facebook executive, D’Angelo is well-known in the industry. When Kevin Systrom launched Instagram and encountered technical issues, he thought: “Who’s, like, the smartest person I know who I can call up?” The answer, as he later told The New York Times, was D’Angelo. But the Quora CEO has also been described as a private, calculated leader by people who’ve worked with him — and one with some prior history in surprise corporate ousters.His position on the board has also raised eyebrows because Quora has been in increasingly direct competition with OpenAI’s best-known service: ChatGPT. From the start, a powerful AI chatbot that can answer questions in seconds risked undercutting at least some of Quora’s pitch to users. Shortly after OpenAI launched ChatGPT a year ago, Quora introduced Poe, a platform that allows people to ask questions from various AI chatbots, including ChatGPT. In late October, Poe introduced an option for developers to monetize the custom bots they build using its tools. The next month, OpenAI announced users would be able to build customized versions of ChatGPT – and make money from their creations in a new GPT store.

The situation is unusual, said Kellie McElhaney, who teaches corporate social responsibility at the University of California, Berkeley’s Haas School of Business. OpenAI’s board was set up as a nonprofit, which generally do not have fiduciary constraints that are as carefully vetted as for-profit entities, McElhaney said. While D’Angelo remains on the board, one of his cohorts left due to concerns about conflicts of interest. Venture capitalist and LinkedIn co-founder Reid Hoffman departed OpenAI’s board earlier this year, citing his growing number of investments in AI companies.

“It feels like a breakdown of trust with multiple parties,” said McElhaney. “In some situations, you say the board took their eye off the ball. Here, nobody knows what that ball is, and where it’s going.” Altman, in a post on X, said D’Angelo has “always been very clear” about the possible conflict “and doing whatever he needed to do (recusing himself when appropriate and even offering to leave the board if we ever thought it was necessary) to appropriately manage this situation and to avoid conflicted decision-making.” Quora is a big OpenAI customer and “we found it helpful to have customer representation on our board,” Altman went on to say in the post, which came after this story was initially published.In a statement, a Quora spokesperson said Poe is “a neutral platform.” The service “provides consumers around the world with access to AI models from OpenAI, Anthropic, Google, Meta, and many other developers,” the spokesperson said. “Quora is not in the business of training these models ourselves; our role is to enable those who do train models to reach a large audience.”

D’Angelo has said little publicly since Altman’s ousting, but as rumors flared about his conflicts and motives, he reshared a post on X, formerly Twitter, from Replit CEO Amjad Masad. “Have known Adam D’Angelo for many years,” Masad wrote. “Although I have not spoken to him in a while, the idea that he went crazy or is being vindictive over some feature overlap or any of the other rumors seems just wrong.”While Altman’s ouster shocked the industry, it wasn’t D’Angelo’s first experience suddenly removing the leader of a company. In 2012, his co-founder at Quora, Charlie Cheever, was also pushed out. In response to a question on Quora about Cheever’s employment status, D’Angelo said, “We decided it was best” for his co-founder to step away from day-to-day work at the company.

Cheever had little notice or understanding of the breakdown, according to a person familiar with the matter. It was so sudden that employees visited Cheever’s home in tears, asking why he had left. It appeared there was little communication about the decision or why it was made, even to staff, the person said. The co-founders have hardly been in touch since.In the decade since, D’Angelo has continued building Quora into a platform to share knowledge of all kinds – even though it has a long way to go before living up to the lofty prediction of Quora investor Keith Rabois in 2010 that it “will be the most valuable company produced post 2005. Period.” Quora has raised about $300 million from big names in Silicon Valley, according to PitchBook, but nearly 15 years after launching, it has yet to go public or be acquired. The company was valued at $2 billion as of 2019, according to PitchBook.Six years ago, in response to a question on Quora, D’Angelo said he expected AI could help his startup “in all kinds of ways” as the technology “gets more powerful,” including by helping “people write better answers.” But he also suggested all bets may be off if and when AI reaches a point where it “can do anything a human can do.”

“I think some form of knowledge sharing will be important in that world, but AI safety is a much bigger concern,” he wrote. “I think it’s good that some people are thinking about the safety issues now, but I don’t personally think it’s constructive for me to worry about that world as opposed to just adapting as it gets closer.”

The next year, he joined the board of OpenAI.

(Adds comment on X from Sam Altman regarding D’Angelo)

©2023 Bloomberg L.P.