OpenAI, the company behind ChatGPT, had a wild weekend. On Friday, founder and CEO Sam Altman was fired from the board, sparking an employee revolt that is still ongoing. The company now has three CEOs in as many days. A shocking shakeup at one of the most important companies driving artificial intelligence research could have profound implications for how the technology continues to evolve. For better or worse, OpenAI has always maintained that it works for the good of humanity, not for profit — with this weekend’s drama, many AI researchers may find themselves in private companies answerable only to shareholders, not to the public. Things are still changing fast, but here’s what we know so far and how things might play out.
[ Related: A simple guide to the expansive world of artificial intelligence ]
“Too Far, Too Fast”
November should have been a great month for OpenAI. On November 6, the company hosted its first developer conference where it unveiled GPT-4 Turbo, its latest large language model (LLM), and GPT, custom chatbots based on ChatGPT that can be trained to perform specific tasks. While OpenAI is best known for text-based ChatGPT and DALL·E, an AI-powered image generator, the company’s ambitions include developing artificial general intelligence where a computer matches or exceeds human capabilities. Currently, the industry is still debating the broad definition of AGI, and OpenAI is playing a big role in that conversation. This upheaval has the potential to resonate far beyond the company’s own hierarchy.
[ Related: What happens if AI grows smarter than humans? The answer worries scientists. ]
The recent turmoil stems from OpenAI’s complex corporate structure, which was intended to ensure that OpenAI develops artificial intelligence that “benefits all of humanity,” rather than allowing the desire for profit to enable technology that could potentially harm us. The AI venture started as a non-profit in 2015, but later created a for-profit company in 2019 to take on outside investment, including a massive deal with Microsoft. The oddity is that the non-profit organization’s board of directors still has full control over the non-profit company, and they are all prohibited from having a financial interest in OpenAI.
But the six-member board of directors had unfettered power to oust Altman — which it did late last week, to the surprise of almost everyone, including big investors. Microsoft CEO Satya Nadella was reportedly “blindsided” and “furious” by how Altman was fired, as were many OpenAI employees who took to Twitter/X to post heartfelt emoticons in support of Altman.
The board initially claimed Altman was fired because “he was not consistently forthright in his communications,” but the accounts site later differed on the speed and safety of the way OpenAI’s research was commercialized. According to The Information, Ilya Sutzkever, the company’s chief scientist and board member, told an emergency all-hands meeting, “This was the board fulfilling its duty to the nonprofit’s mission, which is to ensure that OpenAI builds [artificial general intelligence] it benefits all mankind.” Sutzkever apparently thinks that Altman is “pushing too far, too fast,” and convinced the board to fire him, with chief technology officer Meera Murati taking over as interim CEO. According to The Atlantic, the problems stem from the pace at which ChatGPT has been rolled out over the past year. The chatbot initially served as a “low-level research preview,” but it exploded in popularity, and with it, features were rolled out faster than the more cautious board members were comfortable with.
In addition to Altman, board president Greg Brockman resigned in protest, which really set off a chaotic weekend.
Three CEOs in three days and a threat of eviction
After an internal pushback from employees, Altman was in talks over the weekend to resume his role as CEO. The extended will-they-won’t-they ultimately failed. To make matters more dramatic, Muratti was then replaced as CEO by Emmett Shear, co-founder of streaming site Twitch, bringing the company to three CEOs in three days. Shear reportedly believes AI has anywhere between a five percent and 50 percent chance of destroying human life, and advocates slowing the pace of its development, which is consistent with the boards’ reported views.
Of course, as one of the biggest names in AI, Altman hit the ground running – both he and Brockman have already joined Microsoft, one of OpenAI’s biggest partners. On Twitter/X Late last night, Microsoft CEO Satya Nadella announced that he was “extremely excited to share the news that Sam Altman and Greg Brockman, along with colleagues, will join Microsoft to lead a new research team on advanced AI.”
As of this morning, more than 500 of OpenAI’s 750 employees have signed on an open letter demanding that the board step down and Altman to be reinstated as CEO. If they don’t, Microsoft has apparently assured them that there are openings for every OpenAI employee. Shockingly, even Sutzkever signed the letter and also posted on Twitter/X that he regretted his “participation in board actions.”
Violent consequences
As of now, things are still evolving. Unless something radical changes in OpenAI, it looks like Microsoft has made an impressive move. Not only does the company continue to have access to OpenAI’s R&D, but it suddenly has its own advanced AI research department. If OpenAI employees pull out, Microsoft will essentially have acquired the $86 billion company for free.
Whatever happens, we have just witnessed a dramatic shift in the AI industry. Despite all the chaos of the past few days, the non-profit OpenAI was founded with laudable goals, and the board seems to have felt seriously that their role was to ensure that AI — specifically artificial general intelligence, or AGI — was developed safely. With an AI proponent like Altman now working for a for-profit company unconstrained by such a lofty charter, who’s to say it will?
Similarly, the reliability of OpenAI is in serious doubt. Whatever its bylaws say, if the majority of employees want to move forward with AGI development, it has a serious problem on its hands. Either the board will have to fire a lot more people (or let them go to Microsoft) and completely overhaul itself, or it will bow to the pressure and change its trajectory. And even if Altman somehow rejoins OpenAI, which seems increasingly unlikely, it’s hard to imagine how the nonprofit’s full control of the for-profit company remains in place. In some ways, the trajectory of AI seems significantly less predictable than it was just a week ago.
Update on November 20, 2023, 2:11 p.m.: Shear, OpenAI’s current CEO, said he would launch an independent investigation into the circumstances surrounding Altman’s firing. While it may be too little, too late for some employees, he says the investigation will allow him to “drive changes across the organization” up to and including “significant management changes.”