Last month, as the tech industry was buzzing about ChatGPT, the research arm of the Defense Department put out an artificial intelligence announcement of its own: An AI bot had successfully flown an F-16 fighter jet in the skies above Southern California.
The news got relatively little attention, but it revealed an overlooked truth: The race to develop the next generation of AI isn’t just between tech companies like Microsoft and Google — it’s also between nations, which are working furiously to foster and develop their own technology.
An international competition over AI technology is playing out at a time of high tensions between the U.S. and China, and some experts said they fear how high the stakes have gotten.
“If the democratic side is not in the lead on the technology, and authoritarians get ahead, we put the whole of democracy and human rights at risk,” said Eileen Donahoe, a former U.S. ambassador to the U.N. Human Rights Council and now executive director of Stanford University’s Global Digital Policy Incubator.
AI has become increasingly intertwined with U.S. geopolitical strategy even as chatbots, digital artwork and other consumer uses are stealing the headlines. What’s at stake is a host of tools that countries hope to wield in a fight for global supremacy, according to current and former U.S. government officials and outside analysts.
And it’s not just about military weapons like autonomous fighter jets. Some of the same advances that are powering ChatGPT may be useful for such varied geopolitical tools as large-scale propaganda machines, new kinds of cyberattacks and “synthetic biology” that could be important for economic growth.
“Within the technical community and some parts of the policy community, this race has been going on for quite some time,” said Jason Matheny, CEO of the Rand Corp., a nonprofit that provides research assistance to the U.S. government.
“But what’s different now,” he added, “is that this is a topic of conversation among the general public. There’s millions of people now who’ve interacted with a large language model” — specifically, ChatGPT and its cousin on Microsoft’s Bing search engine.
On the surface, chatbots may not have much in common with autonomous weapons, but they’re built on similar ideas. AI technology is made up of a series of separate advances going on in parallel including new specialized microchips and a new computing architecture called a “transformer” that Google engineers developed. The “T” in ChatGPT stands for transformer.
One casualty so far is the exchange of technology across borders, similar to the way the internet itself has splintered into competing factions. China’s regulators have told Chinese companies not to offer access to ChatGPT services, Nikkei Asia reported last month, and the Biden administration has tightened controls on the export of AI-related technologies to China.
From the Chinese perspective, the competition has resulted in a “decoupling” that hurts both countries but China even more so, according to a report earlier this year from academics at the elite Peking University. The report was later taken offline, the Hong Kong-based South China Morning Post reported.
But in response to U.S. export controls, Chinese President Xi Jinping has emphasized a goal of technological self-reliance.
AI dominance isn’t necessarily winner-take-all. China does more with facial recognition tech than other countries do, using it as a form of control, but censorship may hold it back in the area of large language models.
Matheny said that for the U.S. to maintain an edge, it has to look at several essential components: computing power with microchips, large amounts of data, advanced algorithms and talented engineers.
“Each of these is sort of a strategic resource,” he said. “There’s not an endless supply of people who have the expertise needed to build these large AI models.”
To make the race even more complicated, the biggest source of advanced chips is Taiwan, the island that China claims as its own.
“It’s an inconvenient feature of geography that one of the most important parts of the AI supply chain is also one of the most complicated places geopolitically, 100 miles from mainland China,” Matheny said.
Both the U.S. and China have committed vast resources to AI development. The Defense Department is spending $1.5 billion over five years on AI, and last year Congress added another $200 million. The Defense Advanced Research Projects Agency, or DARPA, which tested the F-16 jet, has separately said it was spending billions of dollars. China’s spending is less clear, but estimates are in the billions of dollars.
In the private sector, the U.S. and China are Nos. 1 and 2 for total private investment in AI, with U.S. investment three times higher than China’s, according to a 2022 report by Stanford University.
“It’s not just about what AI gets invented. It’s about who applies it first,” said Christopher Kirchhoff, a former director of strategic planning for the National Security Council who helped lead the Pentagon’s Silicon Valley office, in an email.
Jake Sullivan, the Biden administration’s national security adviser, has underscored how important AI capabilities are in the eyes of the White House. In what he called a strategic shift, Sullivan said in a speech last year that it was no longer enough for the U.S. to be ahead of other nations on AI but instead “must maintain as large of a lead as possible.”
The competition has most of the elements of a new arms race, analysts said, with all the terrifying scenarios, big budgets and international maneuvering that the phrase entails.
Calls for de-escalation — and even a treaty — are growing louder.
“This is Cold War logic all over again,” said Wendell Wallach, the co-director of an AI program at the Carnegie Council for Ethics in International Affairs.
“Are we ratcheting up the tensions between ourselves and China to the point where we’re putting ourselves in a trap?” he asked.
Last month, the Dutch and South Korean governments co-hosted what they said was the first global summit on “responsible” applications of AI in warfare, and more than 50 participating countries including the U.S. and China endorsed a nonbinding statement on “the need to put the responsible use of AI higher on the political agenda.”
Also at the summit, the Biden administration proposed a set of ideas to keep AI weapons under control, such as one proposal that deadly arms be “capable of being deactivated if they demonstrate unintended behavior.”
A week later, Costa Rica hosted a regional conference on the same subject, demonstrating how widespread the concerns are.
AI is now so tied up in international affairs that it’s become a fixation lately for Henry Kissinger, the 99-year-old former secretary of state. At an event last year, he called on the U.S. and China to begin negotiating limits of some kind, because without them, he said, “it is simply a mad race for some catastrophe.”
Other countries besides the U.S. and China seem to believe that if they’re not competitive on AI, their security will be at risk.
“The one who becomes the leader in this sphere will be the ruler of the world,” Russian President Vladimir Putin told a group of students in 2017. The next year, Russia said it was testing a semi-autonomous tank in Syria, though it got poor reviews, and in Ukraine, both Ukrainians and Russians are pursuing autonomous drone technology, Wired magazine reported.
ChatGPT has shown how easy it may become for a country to create persuasive propaganda on a large scale and ship it abroad, potentially accelerating a conflict, said Joe Wang, a senior director for foreign policy at the Special Competitive Studies Project, a nonprofit set up by former Google CEO Eric Schmidt to “strengthen America’s long-term competitiveness.”
And the potential for other applications has no clear ceiling yet.
“We are at the beginning of the beginning, in terms of a new era of not just strategic competition but how a new technology is changing the landscape of literally everything,” said Wang, a former official at the State Department and the National Security Council.