China poaches U.S. AI talent from OpenAI as it eyes Claude, ChatGPT
Tencent’s Chief AI Scientist Yao Shunyu (right) discusses the technology’s outlook with Tencent
CNBC | Evelyn Cheng
BEIJING — A former OpenAI researcher is now chief AI scientist Tencent He wants to create artificial general intelligence in China.
This signals a shift in the US-China technology race.
Artificial intelligence (AGI) with human-level or higher capabilities has long been a goal of US companies such as OpenAI, Anthropic and Alphabet, which acquired British startup DeepMind.
In their rush to catch up on artificial intelligence and faced with U.S. chip controls, Chinese companies have instead focused on ways to use the technology in applications ranging from factories to consumer electronics. Baidu CEO Robin Li had previously predicted that it would take until at least 2034 to reach AGI, as opposed to Elon Musk’s 2026 prediction at the time.
But as Chinese companies recruit talent from Silicon Valley, they are bringing their vision of the United States with them.
“My personal goal is that we should establish a long-term AGI organization in China,” said Tencent Chief Artificial Intelligence Scientist Yao Shunyu, who joined the company last year after leaving OpenAI, in his speech translated from Mandarin by CNBC.
Yao was onstage discussing the next phase of AI development with Tencent’s Cloud executive Dowson Tong on Friday at an event the company hosted in Beijing in partnership with local officials. A senior official from Beijing gave the opening speech.
The AGI vision will require fundamental knowledge, products and frontier research, Yao said Friday.
Yao said that the untapped potential is in the “trillions of dollars” level and said, “I don’t think ChatGPT or Claude will be the only super application.” The performance of an AI tool is most important, followed by cost, he said, adding that the way forward in China is through smaller AI models and more consistent performance on key tasks.
His optimism contrasts with growing wariness about artificial intelligence in the US
anthropic warned on Thursday Frontier models are approaching the point where they can evolve without human supervision. The company has called for the industry to slow down or pause in new model development to avoid disrupting society.
The San Francisco-based startup called on Washington to take action earlier this year. The USA maintains its leadership over Chinese models. Anthropic has emphasized AI security since its founding and has faced criticism from rivals that its security warnings are designed to stifle competition.
