December 6, 2023
More cargo ships from Ukraine use civilian corridor despite Russian threats

“The mission continues,” Sam Altman, co-founder of OpenAI, the startup behind ChatGPIT, tweeted on Nov. 19. But exactly where this will continue is unclear. Mr. Altman’s tweet was part of the announcement that he was joining Microsoft. Two days earlier, surprising Silicon Valley, he was fired from OpenAI for not being “consistently candid in his communications with the board.” Then Microsoft boss Satya Nadella announced that Mr Altman would “lead a new advanced AI.” [artificial intelligence] Research team within the tech giant”. At first it seemed that Mr. Altman would be accompanied by some of his former colleagues. Many more may follow. The majority of OpenAI’s 770 employees have signed a letter threatening to resign if the board fails to reinstate Mr. Altman.

The fraud involving the world’s most popular startup is far from over. The Verge, a tech-focused online publication, has reported that Mr. Altman may be willing to return to OpenAI if the board members responsible for his dismissal themselves resign. Mr Nadella also seems to accept that possibility. His maneuver may seem clever either way. If Mr Altman returns, Microsoft, OpenAI’s biggest investor, will have backed him through the crisis, cementing an important corporate relationship. If Mr. Altman and his friends join Microsoft, Mr. Nadella could look even smarter. He will bring home the talent and technology on which the world’s second most valuable company is staking its future.

Microsoft has long invested in various forms of AI. It first announced it was working with OpenAI in 2016, and has since invested $13bn for a 49% stake in the startup. The deal means OpenEye’s technology will have to run on Microsoft’s cloud-computing arm Azure. In return OpenAI has access to Microsoft’s massive amounts of processing power, which it needs to “train” its powerful models.

The investment became significant for Microsoft with the launch of ChatGPT a year ago. Chatbots became the fastest-growing consumer software application in history, reaching 100 million users in two months. Since then, Microsoft has been busy working out how to incorporate the startup’s technology into its own software. It has launched ChatGPT-like bots to go along with many of its offerings, including productivity tools like Word and Excel; Bing, its search engine; And even its Windows operating system.

Bringing parts of OpenAI in-house would be a smart move. Technology is at the heart of Microsoft’s future. Having direct control over it eliminates the risk that OpenAI could take its technology in a different direction. And such an effect will be achieved for a bargain. Before being fired, Mr. Altman was hoping to raise new funding for OpenAI, which would value the company at about $86 billion. Hiring OpenAI’s boffins in this way is something that will be more difficult for antitrust regulators to challenge than an outright acquisition. Investors seem curious. Microsoft’s share price fell slightly on the news of Mr. Altman’s dismissal. That loss was reversed when their new program was announced.

Still, there will be risks in this step. Is a prestigious one. A pillar of Microsoft’s AI strategy is to keep the technologies away from each other, thus protecting it from any embarrassment that might occur if ChatGPT goes awry. In contrast, when Facebook’s parent company Meta released its science AI chatbot Galactica, the tool began building research. The public response was so critical that Meta removed it.

Some analysts believe that Microsoft will no longer need to insulate. It has invested heavily in managing AI risks, with teams working on issues including security, privacy and limiting inappropriate behaviour. Mark Moerdler of broker Bernstein says the Microsoft version of OpenAI’s GPT model comes with more guardrails than the startup’s. The company’s launch of its own range of products like ChatGPT shows it is confident it can manage some of the reputational criticism.

Mr Moerdler argues that a bigger risk is that taking OpenAI in-house could lead to a “short-term slowdown in the progress of the technology”. It will take time to get a team led by Mr. Altman off the ground within Microsoft because new models need to be designed and trained. If OpenAI lost its most talented employees in the meantime, it could slow the development of its new products, which Microsoft still relies on to improve its software.

The third danger is that OpenAI’s talent goes somewhere else entirely, not to Microsoft. Marc Benioff, the boss of Salesforce, another software company, has said he would hire any OpenAI researchers who resign.

Whether they will run elsewhere will depend in part on the precise arrangement of Mr. Altman’s new outfit. Early indications suggest it may get considerable independence. Mr. Nadella called Mr. Altman the “CEO” of the new entity. Barry Briggs of Directions on Microsoft, a consultancy, points out that Microsoft has tried to give autonomy to its new acquisitions in the past, citing the experiences of LinkedIn and GitHub in 2016 and 2018.

This time the stakes are much higher: OpenAI’s talent is in high demand and its technology is key to Microsoft’s future. Mr. Nadella will hope to have secured his firm’s interests, whether Mr. Altman takes his new job or returns to the startup he founded. But the chaos is not over yet.

© 2023 The Economist Newspaper Limited. All rights reserved.

From The Economist, published under license. Original content can be found here


Leave a Reply

Your email address will not be published. Required fields are marked *