It was no coincidence that OpenAI’s board abruptly fired CEO Sam Altman and repatriated him days later.

Indeed, this boardroom drama represented the boiling over of tensions that had long been smoldering beneath the company’s surface.

After a tumultuous few days, Altman is back at the helm of the company, with a newly formed board charting the path forward, but he believes the cause of OpenAI’s turmoil lies in the company’s unusual structure. It will be done.

OpenAI was founded in 2015, by Altman, Elon Musk, and others as a nonprofit research organization. It was like the anti-Big Tech companies. It would put principles before profits. As OpenAI said at the time, the company wanted to develop AI tools that would “benefit humanity as a whole, without being constrained by the need to generate economic profit.”

But in 2018 two things happened to him. First, after Musk announced that he had invested $50 million, he resigned from OpenAI’s board of directors, cutting the then-unknown company off from the entrepreneur’s significant financial backing.

Second, OpenAI executives increasingly recognize that developing and maintaining advanced artificial intelligence models requires vast computing power and is incredibly expensive. became.

Balancing Sam Altman’s ideals with the need for funding

A year after Musk retired, OpenAI launched a for-profit arm. Technically, this is a so-called “profit cap” unit, meaning that the investor’s potential profit is limited to a certain amount. Any remaining money is reinvested into the company.

Still, the nonprofit’s board and mission continued to govern the company, creating two competing tribes within OpenAI. These are adherents to the “serving humanity, not shareholders” credo and those who embrace Silicon Valley’s more traditional approach to investor money. We aim to bring our products to the world as quickly as possible, conquer the market, and become the industry leader.

Altman, a 38-year-old techno-optimist who previously ran the famous startup accelerator Y Combinator, sought to combine the two approaches. He chose a middle path, gradually introducing his new OpenAI tool first to small groups and then to larger groups, refining the tool before releasing it publicly.

ChatGPT’s success attracts huge tech funding

Last year, when OpenAI disrupted the technology industry with the launch of ChatGPT, the company’s most prominent investor, Microsoft, significantly increased its stake. Increased commitment to OpenAI by $13 billion.

Microsoft became the financial engine driving OpenAI, but the nonprofit’s board still made decisions. Despite Microsoft’s significant investment, the company did not have a seat on its OpenAI board.

All of this led to Mr. Altman’s sudden ouster from the company earlier this month.

The board itself has not yet explained why it fired Mr. Altman but has only vaguely stated that it believes Mr. Altman “has not consistently and candidly communicated with the board.” is. And the structure of the company gives the board that right, giving it unlimited and unrestricted power to fire the CEO whenever it sees fit.

Before Altman’s dismissal, he had clashed with board members in his rush to commercialize OpenAI products, according to people involved in the discussions. Board members are wondering whether Altman is considering the risks of AI products seriously enough or whether he is simply trying to maintain the company’s dominant position in the crowded and competitive world of generative AI development. I was concerned.

The dangers of powerful AI range from the accelerated spread of disinformation to mass job losses, and the exploitation of human identity by bad actors.

The question was whether Mr. Altman would abandon his OpenAI founding principles and try to grow the company and acquire customers as quickly as possible. If so, is he unfit to lead a nonprofit organization dedicated to developing AI products with “no financial obligations”?

Whatever the reason, the board wants to fire Mr. Altman. When I thought about it, there was nothing Microsoft or company executives could do. This dramatic gesture and subsequent reversal highlighted the tension at the heart of OpenAI’s structure.

An anonymous letter from a former OpenAI employee during the Altman drama asks the board to investigate whether Altman prioritized commercial products and fundraising goals over the nonprofit’s founding mission.

“We ask the Board of Directors to remain committed to OpenAI’s original mission and not succumb to profit-driven pressures,” the letter reads. “The future of artificial intelligence and the well-being of humanity depends on your unwavering commitment to ethical leadership and transparency.”

Uneasy Solution

OpenAI’s board initially refused to consider the possibility of Mr. Altman’s return, but then something happened that couldn’t be ignored. 702 of OpenAI’s 770 employees have pledged to leave if Altman does not return. Employees also requested that new circuit boards be installed. That was the case, and Mr. Altman quickly returned to his CEO role.

The only member of the new temporary board is his former director. This is Adam DiAngelo, the CEO of the question-and-answer site Quora. He had voted to expel Mr. Altman.

People familiar with Silicon Valley executives sat next to him. They include Brett Taylor, a longtime Silicon Valley executive and former CEO of Twitter, and former Treasury Secretary Larry Summers.

OpenAI’s current charter states that it is committed to developing artificial general intelligence, also known as AGI, a type of AI superintelligence that can outperform humans and “does no harm to humanity or illegitimate powers.” It is listed.

But success in Silicon Valley almost always requires enormous scale and concentration of power. Thanks to that, Microsoft, OpenAI’s biggest supporter, has been able to become one of the most valuable companies in the world. It’s hard to imagine Microsoft investing $13 billion in a company without thinking that one day it won’t have a permanent place in the industry.

According to the current mandate of the Board, the development of AI systems should be carried out with the main goal of helping all humanity, without considering the benefits to external investors.

Nevertheless, the for-profit company OpenAI continues to recruit wealthy enthusiasts who want to join the AI ​​gold rush. The two are in conflict and there is no clear possibility of them living together.

The new board is expected to grow and include representatives from Microsoft. One of the Board’s tasks is to focus on the structure of OpenAI. Do hybrid models have too much friction? Or is there a way to move forward at the halfway point?