Photo Credit: Financial Times
By: Daniel Zhang, Columnist
In the past five years, there has been an explosive increase in the use and research of artificial intelligence in emerging military technologies. A faceoff between the United States and China is almost guaranteed and some even suggest a forthcoming “AI arms race” between two great powers.[i] However, the concept of “AI arms race” oversimplifies the upcoming AI revolution. “AI” itself is an ambiguous term while “arms race” does not capture the entirety of what’s coming next. Framing AI developments as a new Cold War era arms race risks overlooking the complexities of AI technology, the non-military impact of AI, and the need for private-public partnerships in national security.
The disruptive potential of AI is real, and US and China are actively competing in researching the revolutionary implications of AI for defense. In 2017, China released the “New Generation Artificial Intelligence Development Plan,” which stated its aim to lead the world in AI by 2030.[ii] While the Pentagon continued to advance AI initiatives with the establishment of the Joint Artificial Intelligence Center (JAIC) in 2018,[iii] the US government struggles to come up with a comprehensive national AI strategy.[iv] To wisely counter the threats in China’s AI plan, the US must focus on advancing research on specific technologies and analyze their broader impact outside the military realm.
To start with, artificial intelligence is not a single technology but a wide range of techniques with various applications. Without detailing the exact use and abilities of specific AI applications in military, the abuse of the term – especially in media – could be misleading. AI is not a weapon despite narratives such as “killer robot” and “autonomous machine” in Hollywood movies. In the foreseeable future, the application of AI in the military spans from cognitive electronic warfare to video and imagery analysis by computer algorithms like Project Maven.[v] Machine learning, or specifically deep learning, currently tops the list among all AI technologies for defense application. Deep learning techniques, such as the use of deep neural networks and pattern recognition, allow machines to learn by analyzing huge amounts of data. While the Chinese government and military are heavily invested in deep learning[vi] the US is not taking deep learning seriously. Limited steps were taken by the government [vii] or the Pentagon[viii] to explore AI technologies including deep learning as of early 2018, however. It appears that only a single project on deep learning is currently in the works.[ix] An AI expert once commented that nobody in the Pentagon “had a clue how to properly buy, field, and implement” such technologies before Project Maven.[x] Although such a comment is certainly exaggerated to a degree,[xi] it shows that there is much for the US military to catch up on in regards to AI technologies.
It is important to note that even with active research and investment in AI, some of those associated technologies, such as fully autonomous weapons, remain in their early phase.[xii] Much of those so-called “autonomous drones” currently deployed on the battlefield require extensive human involvement.[xiii] That is not to say that we should not consider the implications of their later use, but in reality, these technologies are hardly mature and may not even be eventually deployed on the battlefield due to ethical or political risks. All that is to say, there will not be one exclusively military AI competition between US and China; instead, there will be many – depending on which aspects of AI each country prioritizes and develops for military application.
The term “arms race,” also poorly describes the current innovations in AI and fails to capture the multilevel implications of AI development. By comparing the AI competition between the US and China to the nuclear arms race between the US and the Soviet Union during the Cold War, the term portrays AI as an entirely military-based technology and overlooks essential differences between AI and conventional or nuclear arms. First, AI is mostly about software while nuclear weapons largely emphasize hardware. The implication of that along with its dual-use nature makes it difficult to hide AI research in the dark. Both private and public sectors are heavily involved in AI research and their latest achievements, especially from the private sector, are inevitably displayed on an international level. Nuclear weapon programs, on the other hand, are mostly controlled by military and government who often actively seek to hide – though not always successfully – traces of development.
Unlike making dirty bombs, the development of AI technologies has both civil and military applications. AI could affect national security in far-reaching ways, such as the disruption of employment and economic inequalities in the military and beyond.[xiv] Examining the implications of AI through only the lens of weaponization or arms race risks losing the still unknown economic and political implications.
Moreover, the role of the private sector plays a vital role in developing AI for national defense. The private-public partnership (PPP) needed is on a much larger scale than the military-industrial complex during the Cold War. The most cutting-edge product research on AI is taking place in companies like Google with its famous Alpha Go program and the soon-ending partnership with the Pentagon on Project Maven.[xv] China’s 2030 plan details the military-civil fusion in AI to take advantage of the dual-use applications for national defense. Chinese military’s Central Commission for Integrated Military and Civilian Development (CCIMCD) formed in 2017 also aimed to unify the central power through integrating civilian technologies into the defense industrial base.[xvi]
Alarmingly for the US, the current military-industrial complex lacks the skills to build complex AI systems[xvii] and the future of private-public partnerships look bleak as Google is said to end its Maven contract with the Department of Defense due to employee protest. A similar attempt to the Chinese PPP model is crucial for the U.S. to focus its resources and boost AI research and development.
There is no doubt that a national security strategy on the artificial intelligence is imminent. The competition of AI in the age of great power competition is very much a reality. As a part of the Pentagon’s third offset strategy, AI needs to be evaluated through multiple levels and more than just in the military sense.
[i] Nicholas Thompson and Ian Bremmer, “The AI Cold War That Threatens Us All,” Wired, October 23, 2018, https://www.wired.com/story/ai-cold-war-china-could-doom-us-all/.
[ii] Graham Webster et al., “China’s Plan to ‘Lead’ in AI: Purpose, Prospects, and Problems,” New America (blog), accessed February 6, 2019, https://www.newamerica.org/cybersecurity-initiative/blog/chinas-plan-lead-ai-purpose-prospects-and-problems/.
[iii] “DOD AI Industry Day Fosters Relationships With Industry, Academia,” U.S. DEPARTMENT OF DEFENSE, accessed February 9, 2019, https://dod.defense.gov/News/Article/Article/1701496/dod-ai-industry-day-fosters-relationships-with-industry-academia/.
[iv] Will Knight, “Here’s How the US Needs to Prepare for the Age of Artificial Intelligence,” MIT Technology Review, accessed February 6, 2019, https://www.technologyreview.com/s/610379/heres-how-the-us-needs-to-prepare-for-the-age-of-artificial-intelligence/.
[v] Zachary Fryer-Biggs, “Inside the Pentagon’s Plan to Win Over Silicon Valley,” Wired, December 21, 2018, https://www.wired.com/story/inside-the-pentagons-plan-to-win-over-silicon-valleys-ai-experts/.
[vi] Guest Blogger for Net Politics, “China’s Artificial Intelligence Strategy Poses a Credible Threat to U.S. Tech Leadership,” Council on Foreign Relations (blog), December 4, 2017, https://www.cfr.org/blog/chinas-artificial-intelligence-strategy-poses-credible-threat-us-tech-leadership.
[vii] Carlos E. Perez, “Is It Time to Panic about American Ignorance of Deep Learning?,” Medium (blog), March 17, 2018, https://medium.com/intuitionmachine/is-it-time-to-panic-about-american-ignorance-of-deep-learning-b8a7bd1641c9.
[viii] Carlos E. Perez, “The US Military Needs to Urgently Rethink Its Deep Learning Strategy,” Medium (blog), March 18, 2018, https://medium.com/intuitionmachine/the-us-military-is-dangerously-behind-with-deep-learning-2d929929e595.
[x] Kate Conger and Dell Cameron, “Google Is Helping the Pentagon Build AI for Drones,” GIZMODO (blog), March 6, 2018, https://gizmodo.com/google-is-helping-the-pentagon-build-ai-for-drones-1823464533.
[xi] Perez, “The US Military Needs to Urgently Rethink Its Deep Learning Strategy.”
[xii] Kenneth1 Anderson email@example.com, “Why the Hurry to Regulate Autonomous Weapon Systems-but Not Cyber-Weapons?,” Temple International & Comparative Law Journal 30, no. 1 (March 2016): 20.
[xiii] Charli Carpenter and Lina Shaikhouni, “Don’t Fear the Reaper,” Foreign Policy (blog), accessed February 6, 2019, https://foreignpolicy.com/2011/06/07/dont-fear-the-reaper/.
[xiv] Darrell M. West, “Will Robots and AI Take Your Job? The Economic and Political Consequences of Automation,” Brookings (blog), April 18, 2018, https://www.brookings.edu/blog/techtank/2018/04/18/will-robots-and-ai-take-your-job-the-economic-and-political-consequences-of-automation/.
[xv] Tom Simonite, “Artificial Intelligence Fuels New Global Arms Race,” Wired, September 8, 2017, https://www.wired.com/story/for-superpowers-artificial-intelligence-fuels-new-global-arms-race/.
[xvi] Leo Lin, “China’s Answer to the US Military-Industrial Complex,” The Diplomat, April 11, 2017, https://thediplomat.com/2017/04/chinas-answer-to-the-us-military-industrial-complex/.
[xvii] Perez, “The US Military Needs to Urgently Rethink Its Deep Learning Strategy.”