Managing Nuclear Risks: The Emerging Technologies Challenge - The Diplomat, May 24, 2019

In the last week's column for The Diplomat, "Managing Nuclear Risks: The Emerging Technologies Challenge", I examined the impact of emerging technologies on nuclear risks that need to be addressed much more directly.

Conversations around nuclear risks and the need for strategic stability have picked up momentum in the run up to the 2020 Nuclear Nonproliferation Treaty (NPT) Review Conference. Nuclear risks are changing for the worse because of geopolitical and technological changes, but the latter, in particular, needs to be addressed much more directly.

As far as the geopolitical dynamics are concerned, the shift in the global balance of power equations has meant that rulemaking and enforcement have become the victims of great power politics because both rulemaking and enforcement require some consensus among powers. But a second significant factor should be noted that could also lead to possible breakdown in nuclear deterrence: The impact of new technologies.

Emerging technologies such as Artificial Intelligence (AI) and hypersonic weapons, among others, are posing new questions to the traditional understanding of stability and deterrence. These emerging technologies are unfortunately not being discussed enough within Track 1 NPT platforms as directly as they should be. It can, however, be argued that these new technologies could increase regional and global nuclear insecurities and therefore it is necessary to discuss this more seriously in order to manage the risks associated with nuclear weapons.

For the full article, click here.



The development of hypersonic weapons could potentially have a big impact on nuclear stability. A key problem here is that such weapons can increase the possibility of surprise attacks because of the difficulty of detecting and countering them. They thus present the possibility of offering a temptation to attack first in the hope of decapitating the enemy leadership or even destroying second-strike nuclear retaliatory capabilities. Even if states do not plan such attacks, the fear that they may do so could drive potential targets to develop riskier nuclear operational plans, such as launch on warning.

Alternatively, it could also spur nations with smaller nuclear arsenals, such as China, to increase the size of their nuclear forces so that at least some survive such a first strike. This could lead to a destabilizing arms race spiral between key great powers, all of whom are working on developing these weapons.

An additional problem is that hypersonic weapons potentially blur the line between conventional and nuclear weapons, which will also increase instability because of the discrimination problem. It is more stable if early warning systems can clearly distinguish between a nuclear attack and a conventional one. If the same system can undertake both conventional strikes and nuclear attacks, a prudent defense system will have to assume that any attack it detects is a nuclear attack because the risks of assuming the alternative will be too high. Though this is a problem that is not confined to hypersonic weapons, the combination of the risk of a first strike combined with the discrimination problem makes the danger particularly acute.

Another evolving technology is Ballistic Missile Defence (BMD), which also has similar problems. Nuclear deterrence and stability are based on mutual vulnerability, and BMDs threaten to undo this key source of stability, with unpredictable consequences. This is not a new problem of course: the fear that one side might develop a strategic technological advantage was one reason for the Cold War U.S.-Soviet ABM Treaty. It might be time to revisit the possibility of developing a wider ABM treaty-like instrument in order to avoid a larger technology arms race as ballistic missile defense technologies slowly improve.

A third area is outer space, which Russia has been raising repeatedly in the NPT platforms. The increasing use of and reliance on space for national security missions has led states to develop counter-space capabilities to disrupt, deny, degrade or destroy space systems. These include both kinetic and non-kinetic means.

Counter-space capabilities such as anti-satellites (ASAT) add an inherently destabilizing dimension to outer space security. But these are also a problem for nuclear stability because nuclear early warning as well as command and control systems depend on space-based assets far more today than a few decades ago. While these are not new capabilities – the United States and Soviet Union had conducted several ASAT tests in the 1970s and 1980s – the greater reliance on space assets increases nuclear vulnerabilities.

The NPT faces myriad problems, of course, and many will be discussed in the coming deliberations. But these emerging technologies represent a danger that should not be minimized in the debates about how much has been achieved, whether all sides have lived up to their end of the NPT bargain (especially nuclear weapons states), and how to proceed forward.

The Trouble With China’s Edge in the AI Arms Race - 10 August 2018

In this week's column for the Diplomat, I write on the AI arms race and possible implications, The Trouble With China’s Edge in the AI Arms Race. Beijing’s rising capabilities, while not surprising, should raise concerns for the region.

Artificial Intelligence or AI, simply described as the machine intelligence, has come to apply itself in several different sectors across countries in recent years, including healthcare, finance, education and security. But it has also increasingly become inserted into wider geopolitical conversations about the capabilities of major powers, including the United States and China.

For the full post, click here.



Within that aspect of the ongoing conversation, in terms of market share within the industry, the leaders in the field include the United States, with around 40 percent of the global market by some accounts, with countries like China, Israel, Germany, Canada and Russia fast catching up.

But market share is just one indicator of capabilities in this realm. Arguably the more important criteria to understand the lead in the domain is possibly the data that one is able to gather. On that count, it is estimated that China is possibly well ahead of the United States, perhaps aided by the fact that privacy laws in many countries, especially democracies, can dampen data access, thus hampering the prospects of AI, as noted by Microsoft President Brad Smith recently.

Reports suggest that China is probably ahead of the United States in terms of the amount of personal data collected, owing to many factors including the fact “it is a larger country” and the opportunities for collecting information in a country like China are “colossal.”

Chinese government wants to overtake the United States and be the global leader in the field by 2030. China has been making massive investments to create a huge pool of AI experts – in one indicator, it was reported that during the period 2011-2015, China had published more than 41,000 papers on AI, almost double the number of that of the United States. China is also ahead of the United States in patent applications for AI, with Chinese tech giants including Alibaba and Baidu making massive investments in AI.

The regional competition within China to set up AI hubs is also picking up pace. Tianjin, a coastal province in northern China, for instance, revealed $16 billion in funding for AI in the province. There is a clear economic logic to this too, with China attempting to use the AI as a platform to make the switch from being the world’s factory of cheap goods to a technology pioneer. A Cabinet note in China last year stated that by 2030, it will leap ahead with a global lead on AI theory, technology, and application.

However, there is little doubt that China’s leap in AI will extend far beyond civilian sectors and into the military as well. China intends to apply AI in important security sectors such as in the warfighting domain with special emphasis on naval warfare. China has established many AI institutes, but the most eye-catching one from a security perspective is a new research center established by the PLA Academy of Military Science. China’s National Defense University too has a new center for AI sciences, in addition to Tsinghua University’s “high-end” laboratory for military intelligence.

Both the United States and Russia are making efforts to catch up with China in terms of the number of institutions and centers focused on AI. Both countries will reportedly open their own centers for AI in the near future, although there are significant differences in which the three countries approach AI. For instance, Russian efforts are driven closely by the Kremlin, whereas China seeks to combine the civilian and military utilities of AI in a big way. The United States, at least for the time being, is looking at AI for assistance in areas such as logistics and open-source data analysis.

Clearly, there is a big surge in AI research in China, and it is easy to come to some hasty conclusions about how that might play out. According to Jeffrey Ding, a researcher at the University of Oxford and author of a study on China’s AI strategy, China’s capabilities as on date are “only about half those of the US.” He argues that the United States has an edge in AI when it comes to hardware design, algorithmic research, and overall commercial linkages. China’s edge is possibly in terms of the data they are about to churn through a mix of Chinese “private” companies and government agencies. Some groups estimate that China “will possess about 30 per cent of the world’s data” in about a decade.

For many countries around the world, China’s lead in AI itself is not much of a concern. But how that aids Beijing’s military plans in the future should cause some worry. China’s plans to develop large autonomous robotic submarines, to give just one example, create anxieties. These are believed to “large, smart and relatively low-cost,” unmanned in nature and can traverse the international strategic maritime spaces by 2020s.

The traditional advantage held by the western naval powers such as the United States, and Asian maritime powers such as Japan and India, could possibly be eroded by the new Chinese venture. These submarines could be entrusted with underwater operations including mine placement, reconnaissance, and anti-carrier kamikaze ops aimed at the United States and its allies. Zhuhai in Guangdong province boasts of having the world’s largest testing facility for surface drone boats.

While China does not plan to do away with human operators in submarines totally, these AI-assisted underwater boats will have no humans on board and they will have the self-sufficient capacity to head out, complete their missions, and head back to the base without any command support.

It is worth noting that this could also heighten regional security competition more generally. China’s new venture is unlikely to go ahead without any competition, and this could in fact drive a new competition on the naval front as well. In fact, the United States has plans to develop its own Extra Large Underwater Unmanned Vehicles (XLUUVs), for instance, and Boeing and Lockheed Martin are already on the job.

China is also pushing AI in tank warfare by reportedly converting its old Type 59 Soviet tanks into unmanned vehicles fitted with AI. Reports indicate that back in 2014, China had set up its first dedicated research center for unmanned ground vehicles. While the tanks’ main gun was not visible in the recent China Central Television (CCTV) footage, Indian analysts familiar with tank warfare note that it is only a matter of time.

China’s advances in the military domain using AI are nothing short of remarkable. Clearly, such AI-assisted systems, if they are successfully deployed, give China a huge edge. But this will also likely spur a new technological arms race in the region, giving defense planners one more thing to worry about.