How to Win the A.I. Arms Race

China has access to a bigger facial recognition database, but the United States has advantages too

A screen shows visitors being filmed by AI (Artificial Inteligence) security cameras with facial recognition technology at the 14th China International Exhibition on Public Safety and Security. Photo: NICOLAS ASFOURI/AFP/Getty Images

Experts agree that we’re headed toward a future where global leadership in artificial intelligence will translate into economic and military dominance. Unfortunately, authoritarian regimes, such as China, have inherent advantages in research and development. The training of A.I. systems requires data — lots of it. Big data is the oil of the Digital Age and whoever has the most of it — at the highest quality and at the lowest cost — will have a comparative advantage.

Assembling and using big data sets in developed countries, however, can be complicated, for privacy and legal reasons. For example, the European Union is considering rules giving each individual the right to control how their facial data can be used in facial recognition technology — which will (probably) slow development.

Meanwhile, China is developing its facial recognition software on an unimpeded basis, using a big data set developed from the pictures on hundreds of millions of government-issued ID cards — probably the world’s largest facial ID data set, since everyone in China over the age of 16 must have an ID card.

Researchers use health data sets to train A.I. to predict kidney failure, breast cancer, or other health issues. Privacy concerns will almost certainly slow down this research in the developed world. Do you want Google to know if you’re at higher risk of developing cancer? What would Google do with that information, and to whom might it be sold?

By contrast, an authoritarian regime such as China isn’t going to be slowed down by human rights and privacy concerns when assembling big data sets in health, language, facial recognition, or anything else.

Furthermore, A.I. datasets assembled by private companies in developed countries will likely be treated as proprietary. Assume a U.S.-based A.I. firm successfully builds (at great expense) a data set useful for predicting kidney problems and trains its A.I. to accurately predict kidney failure. The firm will have no monetary incentive to make that proprietary data set available to other firms. But an authoritarian regime could treat data sets as a national resource, to be shared among researchers and entrepreneurs, as directed by the government. As the world’s largest country, by GDP size (on a purchasing power parity basis) and population (about 1.4 billion people), China has a lot of resources.

Already, futurist Amy Webb has identified three Chinese corporations among the nine leading the charge into A.I. — the rest are from the United States. And according to research done by the Allen Institute for Artificial Intelligence, China is nearly equal to (and may soon exceed) the United States for quality and quantity of academic A.I. papers.

Historically, authoritarian regimes haven’t done well with cutting edge technologies.

The United States is currently in a strong position in A.I. and China faces its own unique challenges. In developing A.I., or any technology, retaining and recruiting talent is key. It’s estimated that about 50% of China’s millionaires want to permanently leave China. An even greater weakness is that many of China’s rising generation of scientists also want to leave. In a recent survey, 87% of Chinese students studying in the United States for a PhD said they want to stay after graduation (and, it’s not clear the other 13% intend to return to China). While this might reflect some selection bias since students who wanted out of China are more likely to study in the United States, it does suggest that China is at a disadvantage in the competition for talent. The recent protests in Hong Kong have arguably only increased Chinese researchers’ concerns about freedom of speech and thought.

Historically, authoritarian regimes haven’t done well with cutting edge technologies. Researchers usually prefer an atmosphere of political and academic freedom, and dictatorships are bad at resource allocation. Communist Russia lost the technological Cold War and the race to the moon. During World War II, the United States won the race to develop the atomic bomb, and Nazi Germany lost. On the other hand, China seems to have mastered (so far) the art of being an authoritarian regime that fosters technological innovation.

It is time for the United States, and its allies, to take seriously a Cold War type technology race around A.I. — perhaps investing in creating shared big data sets, or perhaps investing in recruiting the world’s top A.I. scientists. The West won the original Cold War, with a lot of imported scientific talent. So, for example, if top Chinese A.I. researchers want to come to the United States, they should be made welcome.

We have a lot of work to do, but it’s imperative for democracies to retain, and preferably grow, their comparative advantage in A.I. — the alternative would be a world dominated by A.I.-enabled authoritarian states.

Steven Strauss is a visiting professor at Princeton University. Follow him Twitter: @Steven_Strauss or join his mailing list at

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store