OpenAI CEO Sam Altman returned to the company on Wednesday afternoon, Beijing time, after a five-day saga of corporate intrigue that rocked the global tech industry.

However, the most critical question in the case remains unanswered: why was Altman ousted? OpenAI has been tight-lipped, and the explanations that have surfaced so far have been lacking evidence. The public will have to wait for the results of an independent investigation by a third-party law firm.

With the return of Altman, OpenAI's board of directors has also been reorganized. Three of the four board members who voted to fire Altman have resigned, including chief scientist Ilya Sutskever. The new board currently has three members, including two new additions: former U.S. Treasury Secretary Larry Summers and former Salesforce co-CEO Bret Taylor. The future board is expected to have nine members. Microsoft, investment firms, and Altman himself have all expressed interest in serving.

Altman's return to OpenAI is only the "end of the beginning." The debate over the speed and safety of AI development has not been fundamentally resolved. However, Altman's return with the support of nearly all employees will give commercial pragmatism the upper hand at OpenAI.

Mystery unanswered

The OpenAI "court intrigue" drama has unfolded like a roller coaster, and finally ended with Altman and Brockman's return. In reviewing the complex and confusing plot, the most critical factor is still shrouded in mystery: why was Altman kicked out?

From beginning to end, OpenAI's official statement was only one sentence, that Altman "was not being transparent enough in his communication with the board." The board later clarified that Altman was not fired for "misconduct" or "security" reasons, but it was still unclear.

To outsiders, it is clear that there are two opposing factions within OpenAI: one led by Chief Scientist Ilya Sutskever, which hopes to carefully advance AI research and ensure its safety; the other, represented by Altman, hopes to quickly promote the commercialization of AI. It is speculated that Altman's recent launch of customized GPTs and GPT Store, his plans to build OpenAI's commercial ecosystem, and his external fundraising to prepare to create an AI chip and hardware company, have exacerbated the concerns of the academic camp.

In the five days of chaos, the media also continued to release new materials in an attempt to find the "court intrigue" culprit. Among the four (former) directors who expelled Altman and Brockman, Quora CEO Adam D'Angelo was first suspected, as his chatbot platform Poe AI is seen as a competitor to OpenAI's GPTs. However, D'Angelo remained on the new board, and Altman even spent Thanksgiving with him, which seemed to dispel the rumors.

Subsequently, The New York Times reported that former director Helen Toner is an executive at the Georgetown University Center for Security and Emerging Technology. A few months ago, Toner wrote a paper criticizing OpenAI's security measures, and Altman clashed with her over it. However, this thought was quickly overshadowed by another mysterious OpenAI AI breakthrough, "Q*".

According to Reuters and The Information on November 23, Sutskever's team made a breakthrough earlier this year, namely that they can use computer-generated data to train models, rather than only relying on existing data on the internet. This broke down a major barrier to developing the next generation of large models. However, Sutskever was cautious about this technology, and in July he founded the "Super Alignment" team, dedicated to putting 20% of the company's computing resources into AI safety issues. On the other hand, OpenAI's senior researchers Jakub Pachocki and Szymon Sidor used Sutskever's results to build a model called "Q*", which can solve basic math problems, demonstrating another milestone in AI capabilities. Brockman has always wanted to integrate the technology into new products.

On November 16, Altman said at the APEC summit that the company had recently made a technological breakthrough that could be said to "lift the veil of ignorance and push the frontiers of discovery." This statement did not attract attention at the time, but the next day Altman was fired, Brockman resigned in anger, and the two researchers Pachocki and Sidor quickly followed suit. Reuters said that OpenAI researchers sent an internal letter warning of the dangers of the "Q*" project, which became the trigger for Altman's firing.

However, according to The Verge, a company insider denied that the board had received the alleged employee letter, and that Altman's firing was not related to research progress. Everything became more confusing again.

As one of the conditions for Altman's return to OpenAI, both sides agreed to conduct an independent investigation into the firing. It is reported that the investigation is likely to be carried out by a third-party law firm. Even Microsoft CEO Satya Nadella has publicly admitted that he does not know the specific reasons, and the outside world can only wait for the results of the investigation.

Pragmatism prevails

A year ago, OpenAI released ChatGPT. For a while, tech giants and startups scrambled to catch up, starting a race for AI development. Will OpenAI’s turmoil give the whole industry a chance to think calmly, or is it just a fleeting episode?

First of all, from OpenAI’s perspective, the loser of this “palace drama” is the AI cautious faction. After this ordeal, the employees will only be more loyal to Altman and his business vision, and the GPT-5 research that was once slowed down may also speed up. Jathan Sadowski, a sociologist of science at Monash University in Australia, believes that OpenAI may further deviate from its non-profit intention and become more and more like a typical profit-oriented Silicon Valley tech company.

From the perspective of the entire industry, OpenAI’s competitors will see this chaos as a godsend opportunity to attract customers. On November 22, when OpenAI’s turmoil was not over yet, Anthropic, supported by Google and Amazon, launched the latest version of its chatbot, Claude 2.1.

Last weekend, when the news of Altman’s dismissal just came out, Amazon’s cloud service AWS set up a special working group to help Anthropic deal with OpenAI customer inquiries, and the two companies also discussed how to attract Snap, Morgan Stanley and other OpenAI customers. Google’s sales team also took the opportunity to launch marketing campaigns to attract OpenAI customers to switch to its platform. In addition, Microsoft itself has also become a competitor of OpenAI, and many companies are considering using Microsoft’s cloud platform Azure, because it also has OpenAI’s models.

According to The Information, investors still regard OpenAI’s chaos as a phenomenon of a single company, and it will not affect their enthusiasm for investing in AI companies, but they are likely to be more cautious when investing in companies with similar governance structures to OpenAI, and also encourage the invested companies to try model providers and open source large models other than OpenAI.

“For all AI companies, this may herald a shift from academic idealism to more emphasis on business pragmatism. For users of this technology, if lucky, it may bring more competition and more options.” The Economist magazine summarized the significance of OpenAI’s chaos.

Editor: Alexander