China Gains on U.S. in Highly Cited AI Research
China is drawing close to the U.S. in the number of its artificial research papers that are among the top 10% most cited, a measure of impactful research, a new report says
China is rapidly gaining ground on the U.S. in artificial-intelligence research, according to a new report, heightening Beijing’s challenge to its strategic rival in a field deemed critical by the Trump administration.
The Allen Institute for Artificial Intelligence, a Seattle-based nonprofit that conducts research and engineering projects, found that China is drawing close to the U.S. in the number of its AI research papers that are among the top 10% most cited, a measure of the highest-impact research.
About 47% of those papers came from the U.S. in 1982. Only 29% do now, while China’s share has risen to more than 26%, according to the study, set to publish Wednesday. China has for years eclipsed the U.S. in total number of papers published, regardless of citations.
The results, which drew on a database of some two million papers published through 2018, add to a body of research that shows Chinese research challenging and even surpassing the lead that the U.S. built on the strength of its university system and private-sector innovation. The Artificial Intelligence Index, a Stanford University-overseen project that tracks AI data, showed last year that China was significantly ahead of the U.S. in number of papers published annually.
Deep LearningShare of top 10% most-cited AI researchpapersSource: Allen Institute for Artificial Intelligence
%ChinaUS19902000’1001020304050
Artificial intelligence—algorithms that attempt to mimic human cognition—has rapidly advanced in recent years, driven by the accumulation of large amounts of data and advances in computing power to analyze the information. Chatbots are replacing customer-service agents in many industries, while e-commerce sites use AI to predict consumer behavior and target advertising.
As the technology advances, it could transform how entire industries and economies do business, as well as determine the future balance of power politically and militarily. China, which became the world’s second-largest economy in 2010 behind the U.S., has long been seen as trailing the U.S. in AI.
“When you look at the cream of the crop, the top 10% and even the top 1% most-cited papers, I was surprised to see how close they are to us,” said Oren Etzioni, the head of the Allen Institute, created in 2014 by the late Microsoft Corp. co-founder Paul Allen. “Just projecting the line, you see in 2020 they’ll surpass us in the top-10% papers and it looks like in 2022 or 2023 they’ll surpass us in the top 1%.”
Such results underline what some researchers say is the sluggishness of U.S. investment in AI research and the Chinese government’s generous funding for it. In 2017, Beijing unveiled plans to lead the world in AI by 2030, setting a top-down agenda for its educational institutions and companies to which they have responded vigorously.
An official at the White House’s Office of Science and Technology Policy pointed to other studies that showed the U.S. still in a commanding lead over China on many measures of advanced AI capabilities. One study published in 2017 showed the U.S.’s AI talent pool was seven times as large as China’s, for example, and other analyses of academic papers have shown Western countries far ahead of China in high-impact research.
President Trump in February issued an executive order directing federal agencies to prioritize and develop budgets for AI research.
Michael Kratsios, the deputy assistant to Mr. Trump for Technology Policy, said the U.S.’s system was the greatest in the world for driving new technologies like AI, and that U.S. policy prioritizes fundamental AI research for the long-term.
Paul Dabbar, the Department of Energy’s Under Secretary for Science, added that the U.S. had a broad ecosystem of scientific institutions, government programs and tech companies that was hard for other countries to replicate.
“With the combination of all that, clearly we’re in an excellent position,” he said.
Write to Asa Fitch at asa.fitch@wsj.com
China is drawing close to the U.S. in the number of its artificial research papers that are among the top 10% most cited, a measure of impactful research, a new report says
China is rapidly gaining ground on the U.S. in artificial-intelligence research, according to a new report, heightening Beijing’s challenge to its strategic rival in a field deemed critical by the Trump administration.
The Allen Institute for Artificial Intelligence, a Seattle-based nonprofit that conducts research and engineering projects, found that China is drawing close to the U.S. in the number of its AI research papers that are among the top 10% most cited, a measure of the highest-impact research.
About 47% of those papers came from the U.S. in 1982. Only 29% do now, while China’s share has risen to more than 26%, according to the study, set to publish Wednesday. China has for years eclipsed the U.S. in total number of papers published, regardless of citations.
The results, which drew on a database of some two million papers published through 2018, add to a body of research that shows Chinese research challenging and even surpassing the lead that the U.S. built on the strength of its university system and private-sector innovation. The Artificial Intelligence Index, a Stanford University-overseen project that tracks AI data, showed last year that China was significantly ahead of the U.S. in number of papers published annually.
Deep LearningShare of top 10% most-cited AI researchpapersSource: Allen Institute for Artificial Intelligence
%ChinaUS19902000’1001020304050
Artificial intelligence—algorithms that attempt to mimic human cognition—has rapidly advanced in recent years, driven by the accumulation of large amounts of data and advances in computing power to analyze the information. Chatbots are replacing customer-service agents in many industries, while e-commerce sites use AI to predict consumer behavior and target advertising.
As the technology advances, it could transform how entire industries and economies do business, as well as determine the future balance of power politically and militarily. China, which became the world’s second-largest economy in 2010 behind the U.S., has long been seen as trailing the U.S. in AI.
“When you look at the cream of the crop, the top 10% and even the top 1% most-cited papers, I was surprised to see how close they are to us,” said Oren Etzioni, the head of the Allen Institute, created in 2014 by the late Microsoft Corp. co-founder Paul Allen. “Just projecting the line, you see in 2020 they’ll surpass us in the top-10% papers and it looks like in 2022 or 2023 they’ll surpass us in the top 1%.”
Such results underline what some researchers say is the sluggishness of U.S. investment in AI research and the Chinese government’s generous funding for it. In 2017, Beijing unveiled plans to lead the world in AI by 2030, setting a top-down agenda for its educational institutions and companies to which they have responded vigorously.
An official at the White House’s Office of Science and Technology Policy pointed to other studies that showed the U.S. still in a commanding lead over China on many measures of advanced AI capabilities. One study published in 2017 showed the U.S.’s AI talent pool was seven times as large as China’s, for example, and other analyses of academic papers have shown Western countries far ahead of China in high-impact research.
President Trump in February issued an executive order directing federal agencies to prioritize and develop budgets for AI research.
Michael Kratsios, the deputy assistant to Mr. Trump for Technology Policy, said the U.S.’s system was the greatest in the world for driving new technologies like AI, and that U.S. policy prioritizes fundamental AI research for the long-term.
Paul Dabbar, the Department of Energy’s Under Secretary for Science, added that the U.S. had a broad ecosystem of scientific institutions, government programs and tech companies that was hard for other countries to replicate.
“With the combination of all that, clearly we’re in an excellent position,” he said.
Write to Asa Fitch at asa.fitch@wsj.com
