The battle cry is AI. The US fears the Dragon, which is close to achieving supremacy in weapons technology, thanks to the help of the US tech community.
Back in September 2017, when Russian President Vladimir Putin candidly observed that the nation who “becomes the leader [in artificial intelligence] will be the leader of the world,” US defence analysts in the Pentagon knew exactly what he was talking about. Only 3 months prior, Beijing presented its ambitious ‘Next Generation Artificial Intelligence Development Plan, ’to bring China’s AI industry in-line with competitors by 2020, become world-leading in several AI fields by 2025, and eclipse the US as the premier global AI innovation centre by 2030.
To many casual observers, particularly economists and company C-suits, the Chinese AI plan simply fell in line with a host of other national AI strategies that were released over the past 16 months, including those of Canada, Japan, France, India, and the European Commission. What made the Chinese plan a bit different was its ambition to outrace the US. Consequentially, stove piping articles on why China will definitely win, will definitely not win, and could possibly maybe win, were ripe with assumptions and speculations based on quantitative indicators, such as: investment data, education and workforce trajectories, big data collection efforts, number of AI companies, and anecdotes of AI implementations. The problem with such a simplified approach is that there is no single metric that guarantees research success, nor do particular economic indicators mean anything within the context of attaining a specific research goal.
By its very own definition, AI is an umbrella term that encompasses numerous individual research strains, stretching from deep learning and computational social choice to neuromorphic computing, robotics, and simply developing better hardware to support sensing, perception, and object recognition. At its core, AI is a system-of-systems, and thus what matters is not only how advanced each individual research strain is, but also how well they interface with each other to fulfila desired objective. Meaning, if your deep learning algorithm is awesome, but your object recognition is crappy, then your autonomous system is going to under-perform. Similarly, even if you have the best algorithm on the market but you lack a concise deployment strategy to collect appropriate data, your AI system could goentirely off the rails and solve the wrong problem.
Consequentially, at the Pentagon, defence analysts did not see China’s development plan as just another nation trying to compete in the AI space. What they are seeing is another chess piece Beijing put on the board to attain technological supremacy in future warfare. Speaking at this year’s annual CNAS (Center for a new American Security)conference, former US deputy secretary of defence Robert Work for example laid out five mutually reinforcing components which he believes Beijing is leveraging to erode US technological advantage in the military domain: industrial acquisition and technical espionage to leapfrog, an emphasis on disrupting, paralyzing, or destroying US battle networks, doctrine and weapons to attack effectively first, developing exotic weapons systems, and exploiting AI for military superiority.
Some readers willlo ok at Bob Work’s 5-point plan as scaremongering and might even be pressed to argue that China is actually becoming a reliable partner in a multilateral world, is a rising force of stability, and a staunch defender of free trade. European politicians have been making exactly those statements to both defend Chinese Foreign Direct Investment and hedge against the Trump administration’s current foreign and trade policy. But even in Europe, governments are waking up to the reality of government-supported Chinese firms taking over European companies in strategic industrial sectors. Thus if Bob Work is correct, then what does this mean for US policy on AI?
As it currently stands, the US government does not have a national AI plan nor has the Pentagon articulated its long-term AI strategy. However, while the White Househas not even bothered to appoint a science advisor, the Pentagon has decided to pick up the gauntlet. According to recent comments made by Thomas Michelli, acting deputy CIO for cyber security at the DoD, the Pentagon is just weeks away from publishing its first broad AI strategy. Similarly, on July 13, Patrick Shanahan, deputy secretary of defence, signed off on a memo to restructure the offices of the Undersecretary of Defence for Acquisition and Sustainment (A&S) and the Undersecretary of Defence for Research and Engineering (R&E). The memostipulates that R&E will be streamlined with a particular focus on cyber, quantum science, AI/machine learning, microelectronics, and directed energy. A&S meanwhile, will push for a comprehensive defence industrial base policy and is specifically ordered to combat China on the industrial front.
Policy-wise this all sounds nice and sweet, if only the tech sector were on board. As it turns out, an increasing number of US headquartered companies and researchers specializing on AI have taken the extraordinary step to either resign, not renew their contracts with the DoD, or vowed to never participate in the development of battlefield technology due to moral and ethical concerns. Chinese companies on the other hand do not have this luxury of choice. Testifying before the US House Permanent Select Committee on Intelligence on July 19, CNAS Adjunct Fellow Elsa B. Kania for example noted that “Baidu is partnering with the China Electronics Technology Group (CETC), a state-owned defence conglomerate, through the Joint Laboratory for Intelligent Command and Control Technologies, to pursue applications of big data, cloud computing, and artificial intelligence, in military command and information system.”
The part where it really gets problematic, is that US tech companies – while being averse to working with the Pentagon – have apparently no moral or ethical quarrels to engage in joint ventures and strategic partnerships with Chinese companies and universities that have clear linkages to the Chinese military. Google for instance, after facing immense pressure from its employees and outside critics, decided to expire its contract with the DoD to help leverage machine learning in analysing drone footage. Meanwhile, Google’s China AI Centre is happily pursuing plans to cooperate with Tsinghua’s Institute for Artificial Intelligence, even though the university is “deeply and institutionally committed to China’s national strategy of military-civil fusion in AI, including with a high-end laboratory for military intelligence and research funded by the PLA’s Central Military Commission.”
The discrepancy in Google’s approach of what the company and its employees deem ethically right and wrong, has even prompted five US Senators to send a letter to Google CEO Sundar Pichai, expressing that “while we regret that Google did not want to continue a long and fruitful tradition of collaboration with the military and technology companies, we are even more disappointed that Google apparently is more willing to support the Chinese Communist Party than the US military.”
For the Pentagon, the tech race is thus not necessarily about money, talent, or even the number of AI companies. The real fight it has to win first is to conquer the hearts and minds of the US tech community, and their willingness – if not obligation -to help protect the American people and its allies abroad from an authoritarian state bound to reshape the future of warfare.