Good morning. Clay Chandler here, writing from Hong Kong.
On Tuesday, senior officials from the U.S. and China met in Geneva for a first-ever discussion about how they might work together to ensure emerging artificial intelligence technologies don’t become existential risks.
Ahead of the meeting, officials from both countries said they worry counterparts will use AI to sow disinformation, establish military dominance, and put algorithms rather than people in charge of critical national security decisions.
Taiwanese AI pioneer Kai-Fu Lee was among the first to warn that AI and data could tip the global power balance. In his 2018 book, AI Superpowers, Lee fretted that “many compare the ‘AI race’ of today to the space race of the 1960s or, even worse, to the Cold War arms race that created ever more powerful weapons of mass destruction.”
At the time, the conventional wisdom was that China had the edge in the AI arms race; it has more people than the U.S. (hence more data), and the sector had massive government support.
But the U.S. has been the clear AI front-runner since November 2022, when OpenAI released ChatGPT. The large language model chatbot now claims more than 180 million users and is locked in competition with domestic rivals including Google’s Gemini, Meta’s Llama, and Anthropic’s Claude 3.
China’s nearest equivalent to ChatGPT is Ernie Bot, a large language chatbot launched by search giant Baidu last August. Baidu says Ernie has more than 200 million users and is faster than ChatGPT4. Chinese tech giants including Alibaba, SenseTime, and Tencent have rolled out chatbots, too—to mostly mixed reviews.
China has about 260 generative AI startups, according to ITjuzi. The Financial Times recently profiled four firms valued between $1.2 billion and $2.5 billion: Zhipu AI, Moonshot AI, Minimax, and 01.ai (founded by Lee). But China’s generative AI startups lag U.S. counterparts in technological development and fundraising.
One key difference between how the U.S. and China run the AI arms race: Chinese firms must operate within a strict regulatory framework designed to bolster the legitimacy of the state but also achieve other aims, including protecting privacy, guarding against deepfakes, and preventing delivery companies from using algorithms to overwork employees.
That’s a far cry from the U.S., where on Wednesday a bipartisan group of senators led by Chuck Schumer punted on AI safety laws, citing concerns about “stifling innovation.” Instead, the committee called for a $32 billion per year increase on federal spending on AI. Schumer described the increase as “surge emergency funding to cement America’s dominance in AI” including “outcompeting China.”
For now, at least, Americans seem determined to maintain their lead in the global AI arms race even at the risk of other homegrown hazards.
More news below
Clay Chandler
clay.chandler@fortune.com
Follow on LinkedIn
