纽约时报中文网 - 英文原版-英Microsoft Debates What to Do With AI Lab in China

When Microsoft opened an advanced research lab in Beijing in 1998, it was a time of optimism about technology and China.

The company hired hundreds of researchers for the lab, which pioneered Microsoft’s work in speech, image and facial recognition and the kind of artificial intelligence that later gave rise to online chatbots like ChatGPT. The Beijing operation eventually became one of the most important A.I. labs in the world. Bill Gates, Microsoft’s co-founder, called it an opportunity to tap China’s “deep pool of intellectual talent.”

But as tensions between the United States and China have mounted over which nation will lead the world’s technological future, Microsoft’s top leaders — including Satya Nadella, its chief executive, and Brad Smith, its president — have debated what to do with the prized lab for at least the past year, four current and former Microsoft employees said.

The company has faced questions from U.S. officials over whether maintaining a 200-person advanced technologies lab in China is tenable, the people said. Microsoft said it had instituted guardrails at the lab, restricting researchers from politically sensitive work.

The company, which is based in Redmond, Wash., said it had also opened an outpost of the lab in Vancouver, British Columbia, and would move some researchers from China to the location. The outpost is a backup if more researchers need to relocate, two people said. The idea of shutting down or moving the lab has come up, but Microsoft’s leaders support continuing it in China, four people said.

“We are as committed as ever to the lab and the world-class research of this team,” Peter Lee, who leads Microsoft Research, a network of eight labs across the world, said in a statement. Using the lab’s formal name, he added, “There has been no discussion or advocacy to close Microsoft Research Asia, and we look forward to continuing our research agenda.”

The debate at Microsoft stands out because the company is one of the few major U.S. tech firms — alongside Apple and Tesla — to keep a foothold in China. As China nurtured a domestic tech industry and geopolitical tensions increased with the United States, American companies such as Google whittled down their presence there. Facebook and other U.S. social media sites such as X have been blocked in China for years.

LinkedIn, which Microsoft owns, shut down its professional social network in China in 2021, citing growing compliance demands. But Microsoft has maintained its Bing search engine as the only foreign search engine in China, though it is heavily censored, and it offers its Windows operating system, cloud computing and applications for corporate customers there.

Microsoft has debated the lab’s future for several years, five people with knowledge of the situation said. It has become a target of national security concerns amid the rise of A.I. and growing aggression between the United States and China. The hypothetical risks are that China could hack or otherwise infiltrate the lab, or that its researchers could leave Microsoft to join Chinese companies that work closely with the government, the people said.

The Biden administration privately asked Microsoft about the lab while drafting a ban over the past two years on new U.S. investments in companies building sensitive technologies in China that Beijing could use to enhance its military, two people familiar with the conversations said. (The proposed rules, issued in August, are not yet final.)

Senators asked Mr. Smith about Microsoft’s ties to China at a subcommittee hearing on A.I. in September. He said the country accounted for 1.5 percent of Microsoft’s sales, which were $212 billion last fiscal year.

Microsoft faces “a tricky balance,” said Chris Miller, the author of “Chip War,” a book that tracks the geopolitical history of tech. “They need to consider where the trust of the political system is going.”

The White House declined to comment.

Microsoft’s lab in Beijing was born when Mr. Gates appointed Kai-Fu Lee, a Taiwanese-born A.I. researcher, to build the operation. (Dr. Lee later left to join Google and now runs a venture capital firm.)

Researchers at the lab, many of whom were at the top of their field, explored technologies such as speech recognition, computer vision and natural language understanding, which are cornerstones in the development of artificial intelligence. Some of the lab’s researchers left for key positions at Chinese tech giants like Alibaba, Baidu and Tencent or helped found start-ups such as Megvii, a facial recognition company that has contributed to a vast national surveillance system in the country.

In 2018, Microsoft said it had invested more than $1 billion in research and development in China over the previous decade. The technical talent and invention from the Beijing lab underpin a key internal argument for supporting it.

But the lab’s success and prestige also attracted attention in Washington, where the White House has increasingly restricted China’s access to crucial technologies, citing national security.

Microsoft’s leaders have discussed how to manage the tensions. Mr. Gates, who is still in regular touch with company executives and supports global engagement, has long backed the Beijing lab, people with knowledge of the matter said. He traveled to China in June and met President Xi Jinping, who told him that he was “the first American friend I’ve met with this year.”

Microsoft’s technology and research leaders, including Peter Lee and Kevin Scott, the chief technology officer, also support the lab, arguing that it has produced critical technological breakthroughs, two people said. Mr. Smith also backs the lab.

“The lesson of history is that countries succeed when they learn from the world,” Mr. Smith said in a statement. “Guardrails and controls are critical, while engagement remains vital.”

In recent years, Microsoft has limited what projects the researchers in China can work on, people with knowledge of the matter said. Last fall, researchers in China were not allowed on the small teams at Microsoft that had early access to GPT-4, the advanced A.I. system developed by Microsoft’s partner OpenAI, they said.

The lab also has restrictions on work related to quantum computing, facial recognition and synthetic media, Microsoft said. The company also blocks hiring or working with students and researchers from universities affiliated with China’s military, it said.

(The New York Times sued OpenAI and Microsoft last month for copyright infringement over the training of their A.I. systems.)

In the outpost of the lab in Vancouver, researchers can freely have access to critical technologies, including computing power and OpenAI systems needed for cutting-edge research, two people with knowledge of the lab said.

Kate Conger contributed reporting from San Francisco.