TOKYO -- Andrew Marshall, a well-known American strategic thinker, died on March 26 at the age of 97. Marshall shaped long-term U.S. military strategy as director of the Office of Net Assessment at the Department of Defense, but he never stepped into the limelight during his career, which ended with his retirement at 93.
One thing that troubled Marshall toward the end of his life was the possibility that an adversary might overcome hard-won U.S. technological superiority, due to the proliferation of high technology.
That nightmare vision is coming closer to reality as Russia, and especially China, absorbs artificial intelligence technology at breakneck speed and applies it to military use.
At an international conference on AI weapons in Geneva from March 25 to 29, participants called for regulations to prevent drones and robots from making decisions about whom to kill on the battlefield. While the U.S. and Russia opposed the move, China supported it, said a person familiar with the meeting.
Behind the scenes, however, China is integrating AI into its military technology at a speed that could upset the strategic balance in the Asia-Pacific region.
The U.S. has grown increasingly alarmed at China's capabilities, particularly after one of its state-owned companies, in June 2017, successfully operated 119 unmanned aircraft using AI. The Pentagon held the previous record, with 103 aircraft. The U.S. is concerned that swarms of Chinese airborne and underwater drones could sink its precious aircraft carriers in a war between the two countries.
Aircraft carries are big targets, and vulnerable to attack, so the U.S. deploys state-of-the-art submarines, Aegis-equipped ships and fighter jets to protect them. But an attack by, say, more than 100 unmanned aircraft, could prove difficult to repel. In addition, drones and other smart weapons can be produced much more cheaply, and in greater quantities, than huge carriers.
The threat does not just come from the sky. According to the South China Morning Post, the Chinese Academy of Sciences is developing AI-equipped submarines for possible deployment in the early 2020s.
The unmanned subs will be capable of carrying out various missions, such as deploying mines and attacking enemy vessels, the Hong Kong daily said. The main target for the underwater drones would be U.S. forces in the western Pacific and the South China Sea.
China is also building up its AI-based cyberwarfare capabilities, aiming to strike at U.S. satellites and paralyze U.S. ships and troops in the event of a conflict. Some experts believe the Chinese military has a cyberunit that is more than 100,000 strong.
If a war breaks out, China will likely employ a Tom Thumb strategy: It will not directly take on the U.S., which still outmatches it. Rather, it will seek to immobilize U.S. forces by targeting their vulnerabilities.
The strategy, if successful, could destabilize the Asia-Pacific region, which is kept tranquil by Washington and its heretofore invincible aircraft carriers. U.S. carrier battle groups, which include 11 flat-tops are meant to keep enemies at bay -- and to send a message when necessary.
In November 2017, for example, the U.S. sent three carriers to waters near the Korean Peninsula, signaling North Korea that returning to the negotiating table, rather than pursuing military confrontation, would be a good idea.
Pentagon officials are reportedly seriously worried that U.S. military supremacy will be in jeopardy if Chinese AI weapons can neutralize American aircraft carriers.
Meanwhile, Russia's AI-assisted cyberattacks are rattling the U.S. and its European allies.
Washington is by no means standing idly by. The Pentagon announced its AI strategy for the first time on Feb. 12 and is working to speed up deployment of the technology.
Although accurate comparisons are difficult, the U.S. is thought to have an edge over China in the number of AI-based weapons. One estimate points out that the U.S. has more than 7,000 aerial drones, compared with a bit more than 1,000 for China.
But the U.S. position is not unassailable. In a report dated Feb. 12, the Brookings Institution, a U.S. think tank, said China plans to spend a total of $150 billion over the next decade as it tries to become the world's top AI power by 2030. The U.S. annual budget for AI is around $1.1 billion.
In 2018, China filed for 53,345 international patents, including those for AI, trailing the U.S. by just 3,000 filings. It is only a matter of time before China surpasses the U.S. in terms of patents, given the combined efforts of the government and Chinese companies.
China's advances in AI raise security concerns in the Asia-Pacific region, particularly if they increase the risk of a military clash between the U.S. and China.
Because drones and cyberweapons are much less visible than warships and fighter jets, it its difficult to assess their military potential. As the U.S. and China become more leery of each other, a slight misunderstanding could rapidly escalate tensions.
"It is difficult to quantify the strength of AI-based cyber and unmanned systems, and how the U.S.-China military balance will change is unclear," said Satoru Mori, a professor at Hosei University and an expert in military technology. "Accurately assessing the military strength of the U.S. and China will become difficult, raising the risk that they will misjudge each other's capabilities.
In an interview with Nikkei in 2006, Marshall lamented a lack of intellectual effort in the U.S. to understand China.
New technologies -- from gun powder, to aircraft to atomic power -- have enriched people's lives. But they have also made war more deadly. It is time to think deeply about how AI will change the nature of China's military activities, and how best to respond.