Three Interesting Predictions on How the OpenAI Crisis Will Impact the Future of AI

Our preconceptions are obsolete: Where do the recent events leave the AI world?

Edit Tuesday Nov 22nd: Sam Altman is back as OpenAI CEO. A couple of important details:

Changes to the non-profit board: The board was composed of six people before the firing: Sam Altman, Greg Brockman, Ilya Sutskever, Adam D’angelo, Helen Toner, and Tasha McCauley. Now it’s only three members.

Altman and Brockman are out as part of the agreement. Among the members who executed the coup, three are out: Sutskever, Toner, and McCauley. D’angelo stays. Two new members: Bret Taylor and Larry Summers.

The board will likely grow over time, including at least one seat for Microsoft to represent the company’s interests or provide it with some oversight so that there are no more surprises, as Satya Nadella told Kara Swisher on Monday.

Internal investigation: A settlement could only be reached because Altman agreed to an investigation to elucidate the reasons that led to his dismissal on Friday. The board had been divided for months before the ouster — it was not a sudden decision but a very premeditated one. It’s still unclear what was the inflection point or which factors weighted the most, but it seems to have been a compendium of many things, instead of one big reason.

---

I’d usually publish my weekly top picks today. Well, it will have to wait for the next Monday — there’s only one top pick this week. No one knows how the OpenAI Crisis will end but it’s a safe bet to say that it will impact AI — including for researchers, companies, users, and enthusiasts — like nothing else this year. I believe it’s crucial that we figure out what’s exactly going on before we go back to anything else. That’s why I want to share with you a brief chronology of the events that took place during the weekend (Friday-Monday), that I will update as I have access to more information.

I also believe it’s just as crucial that we understand how these events will affect the AI world and the broadest tendencies in the industry and community in the short-term future (I will use only confirmed information for the analysis but take it with a huge grain of salt anyway). In particular, I will focus on these questions: AGI (artificial general intelligence) vs industrial AI, incumbents AI companies vs AI startups, and the question of AI safety and alignment.

An unprecedentedly chaotic weekend

On Friday, OpenAI’s 4-member non-profit board abruptly fired CEO Sam Altman, and, as a result, Chairman Greg Brockman resigned. They appointed Mira Murati as interim CEO. Everyone — including those directly affected — was shocked. The board remained silent.

On Saturday, employees and investors, furious and worried about the company’s future, tried to force the board to undo its yet unexplained drastic actions and reinstate Altman and Brockman to their former roles. The negotiations were led by Microsoft CEO, Satya Nadella. Employees were optimistic Altman would return.

On Sunday, after a tense yet infructuous discussion, and against all odds, the board rejected Altman’s offer (he demanded to get his role back that the board resigned, which would bestow him a higher legal and executive power than before). Standing by its decision, the board removed Murati (who tried to get Altman back from the beginning) from CEO and appointed a new interim CEO, former Twitch CEO Emmett Shear.

On Monday, to save the critical situation after the deal fell apart, Nadella decided to appoint Altman and Brockman as co-leads to a new AI research division within the company. In what appears to be a brilliant business move whose details we still ignore, Nadella has managed not only to keep intact its partnership with OpenAI but to capture all the value that had been apparently lost after the board coup. The company's stock value surged on Monday to an all-time high after dropping on Friday, resulting in a net gain. But things didn't end there.

Soon after the news of Altman and Brockman joining Microsoft, 500 OpenAI employees (out of ~770) signed a letter directed to the non-profit board (the number has been growing steadily and is now at 715, almost 95% of the company). Here’s the last paragraph of the letter:

Your actions have made it obvious that you are incapable of overseeing OpenAl. We are unable to work for or with people that lack competence, judgement and care for our mission and employees. We, the undersigned, may choose to resign from OpenAI and join the newly announced Microsoft subsidiary run by Sam Altman and Greg Brockman. Microsoft has assured us that there are positions for all OpenAl employees at this new subsidiary should we choose to join. We will take this step imminently, unless all current board members resign, and the board appoints two new lead independent directors, such as Bret Taylor and Will Hurd, and reinstates Sam Altman and Greg Brockman.

Strikingly, Ilya Sutskever, who has been described as the primary responsible for the coup against Altman and Brockman (perhaps wrongly accused, we don’t know for sure yet), signed too. An hour earlier he had apologized on Twitter for participating in the board’s actions:

I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I love everything we've built together and I will do everything I can to reunite the company.

With Sutskever’s change of mind and the overwhelming support of the entire OpenAI staff, Altman seems to be trying to return as OpenAI CEO. The Verge has reported that “Altman’s move to Microsoft isn’t a done deal.”

An inflection point that twists our preconceptions

That’s what we know so far. How the story will end is unclear. Has Microsoft effectively acquired OpenAI basically for free? Can Altman go back now that the board has appointed yet another interim CEO, Emmett Shear if the immense employee pressure succeeds? Can the rest of OpenAI, if things follow the current course, keep ChatGPT, the API, and the other OpenAI products working? No one really knows.

This whole tech circus turned Shakespearean play has been extremely weird and certainly unprecedented in the history of Silicon Valley. We shouldn’t dismiss the possibility that stranger things may be still coming. To state the obvious, the AI industry and academic landscapes have radically shifted overnight. Any preconceptions you had about what these companies were trying to do or their projections for 2024 are obsolete (even if some of them may prove to remain true after things settle down). The best advice I can give you is to keep an open mind and adapt accordingly (for most people, nothing will change at all, this is intended for those of you who want or need to follow closely what happens in the AI world).

Here are some questions to ponder that I will explore more in-depth below: Who is leading the race right now? Will OpenAI regain its former glory? Where does all this — OpenAI's board disaster, Microsoft’s new powerful hires, the employees’ resignation threat — leave Google DeepMind, Anthropic, Meta, and the other competitors? Is it all about AI products from here on to the detriment of non-profitable research like interpretability? What about AGI, has the dream died? Have they finally realized that alignment is unlikely? It’s impossible to answer these questions with certainty. Not even Altman or Nadella know. But we can try to shed some light on them with the available information.

As a half blogger-half analyst, I need to clarify my thinking to provide you with interesting and useful insights about the present and future of AI. In order to do that — and to help you with your own perceptions — I want to write down and share a few broad insights that I’ve been pondering over the weekend about what’s next. As I said, it’s all very unstable (who could have predicted that in 72 hours the timeline would change so many times in such profound ways?) but it might still be of value to you. I just ask you to take everything below with a huge grain of salt.