Skip to Content
Artificial intelligence

China may overtake the US with the best AI research in just two years

The number of influential AI research papers coming from China is increasing rapidly, a data analysis shows.
March 13, 2019

White House plans to cut funding for science couldn’t come at a worse time for the country’s ambitions to lead the world in artificial intelligence.

The most detailed analysis of Chinese AI research papers yet suggests that China is gaining on the US more quickly than previously thought.

China’s vibrant tech scene has come up with a number of  recent breakthroughs, and the government has recently launched a major initiative to dominate the development of the technology within a matter of years (see “China’s AI awakening”).

Still, it isn’t easy to measure progress in a broad and complex area of technology like artificial intelligence. Previous studies have shown that China already produces a larger number of research papers mentioning AI terms like “deep learning.” But it has always been difficult to ascertain the quality of that research.

The new study aims to solve that problem. It comes from the Allen Institute for Artificial Intelligence (Ai2), a nonprofit in Seattle created by Microsoft cofounder Paul Allen that is focused on fundamental AI research. The institute previously created a tool, called Semantic Scholar, that uses artificial intelligence to make it easier to search and analyze scientific research papers published online.

Using this tool, Ai2’s researchers examined not just the number of AI research papers coming from China but the quality of those papers—as judged by the number of citations they receive in other work. The study suggests that China will overtake the US in the most-cited 50% of research papers this year, the top 10% of research papers in 2020, and the top 1% by 2025.

The top 50% of AI research papers, by citations.
How Chin and the US share the top 50% of AI research papers, by citations, over time.
Allen INstitute for Artificial Intelligence

“Our economy and security has benefited greatly from the cutting-edge research being home grown in our universities and research institutes,” says Oren Etzioni, CTO of Ai2 and a leading AI researcher.“We need to urgently increase AI research funding, and commit to visas for AI students and experts.”

But there are reasons to be cautious about this research, too.

Kai-Fu Lee, a prominent Chinese AI investor who previously established both Microsoft’s and Google’s outposts in China, says the study may overstate things a bit. “There’s definitely momentum,” he said by phone from Beijing, “but the time horizon is farther out.”

Lee, who is the author of a recent book on AI in China,  AI Superpowers: China, Silicon Valley, and the New World Order, says the US still boasts the vast majority of the world’s most influential scientific thinkers, according to measures of an individual’s citations (the US is also still far ahead of China in terms of “best paper awards” at major conferences). “Had there not been a Geoff Hinton or a Yann LeCun,” he says, referring to two AI researchers based in Canada and the US, respectively, “would there have been deep learning?”

Lee is also hopeful that people don’t lose sight of overall scientific progress when talking about the “AI race” between the US and China.

Etzioni agrees that openly published scientific research can benefit everyone, regardless of its country of origin, but he believes the study should  serve as a wake-up call to the US government. “If we move to second place, will the next Google be founded here or in China?” he says. “It’s not a zero-sum game, but it isn’t a picnic either.”

Deep Dive

Artificial intelligence

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

What’s next for generative video

OpenAI's Sora has raised the bar for AI moviemaking. Here are four things to bear in mind as we wrap our heads around what's coming.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.