Media  Foreign Affairs and National Security  2019.03.05

In the age of AI, think tanks must evolve

the japan times on February 26, 2019

I am not particularly interested in the second U.S.-North Korea summit meeting taking place this week in Vietnam because it's easy to predict that the possible consequences will not be very significant, if not meaningless. Therefore I'll write about something that I deem much more important.

An intellectually stimulating international conference titled "The AI Think Tank Forum" took place last week in Palo Alto, California. The forum focused on the policy implications of artificial intelligence, the issue of AI and governance and the impact of AI on think tanks and policy research.

The Canon Institute for Global Studies (CIGS), where I direct its foreign policy and national security team, not only had the privilege to cosponsor the forum but also to send two of its researchers there. Based on their debriefing, I wish to explain the outcome of this unique and significant landmark conference.

The participants were also significant: Harvard Kennedy School's Belfer Center, Brookings, the Center for a New American Society, Chatham House, the Council on Foreign Relations, and the Center for Strategic and International Studies, to name a few. Also participating were leading think tanks from more than 10 countries including Japan, Canada, France, Germany, Italy, India, Israel, Brazil and South Korea.

Ryohei Hisano, CIGS's AI/big data expert was one of the speakers in the first session and Yuki Tatsumi chaired the session on "Impact of AI on Politics and Governance." I really wanted to join them, but my schedule prevented me from participating, although it might have been better for CIGS that I didn't participate this time.

The organizer's visions were crystal clear. With the innovation in AI technologies, GAFA (Google, Apple, Facebook and Amazon) now can do better policy analyses than the existing conventional think tanks. This means the think tanks need to utilize AI/big data in making meaningful policy recommendations.

As far as the participants are concerned, policy people seem to have outnumbered AI scientists. Under the Chatham House rule, I am not supposed to name specific participants, but American participants appear to have been more concerned about China's more advanced AI technologies posing a threat or danger.

Other participants, on the contrary, seemed to be more interested in engaging China and Russia in the making of international rules to regulate AI-related matters. This difference in attitude will be one of the issues to deal with in the years to come.

The following is a summary of what Hisano and Tatsumi stated in the forum:

CIGS has its own research unit on AI/big data, which is unique because it consists of scholars with expertise in mathematics, system engineering, data science or physics. They try to utilize AI/big data not only for economic or financial issues but also for foreign policy and national security analyses.

Big data analyses, for example, on the supply chains of specific precious minerals are potentially useful in analyzing some ethnic conflicts in Africa.

In addition, CIGS is working hard to maximize its synergy effects by currently making combined research efforts on the following subjects:

■ Analyzing U.S.-China bilateral trade issues by utilizing customs data to assess the impact of escalating trade tensions between the two nations.

■ Utilizing data of real estate prices to analyze the risk of real estate bubbles: CIGS's case study on Japan's bubble economy may be applicable to the analyses on real estate prices in China.

■ Utilizing GPS and other positioning data from various devices (i.e., smartphones) to analyze the movement of people, which may provide valuable information for the solution to ensure "smart" border security.

■ Analyses of SNS activities to assess political divides and activities by various active political groups motivated by ideologies, religion, etc.: This can be a useful tool to identify and locate potential extremist groups.

■ Big data analyses on various list of economic sanctions or investment risks may be effective for identifying corporations or other entities that may be "blacklisted" or in assessing the impact of multi-layered economic sanctions.

After listening to the debriefing from the two CIGS participants, I learned two very important lessons. First, we must seriously deal with the negative aspects of the AI technologies. How to deal with the governance of AI and other state-of-the-art technologies is now a common concern for Western nations. If it is the unfortunate reality we must face, it is imperative for all foreign policy and national security analysts to have at least a minimum basic knowledge about the mechanism and application of AI/big data technologies when they discuss foreign policy.

Second, although this may not have been clearly noticed in Tokyo, the world is paying more attention to Japan regarding AI/big data issues. Recent Japanese initiatives in this field, as proposed by Prime Minister Shinzo Abe at the Davos conference last month, were also highly appreciated in the AI Think Tank Forum.

Calling for global consensus on data governance, Abe stated at the World Economic Forum on Jan. 23 that he "would like Osaka G20 to be long remembered as the summit that started worldwide data governance."

All in all, AI-related issues are so important and serious that think tanks must further develop international coordination and work together to make the best use of AI while not allowing the abuse of such technologies. This was one of the concerns commonly shared by all the participants in the AI Think Tank Forum.

As a researcher at a think tank in Tokyo, I am convinced that this is the direction that the leading think tanks in the world must head. This is easy to say, however, but difficult to deliver. For example, there is little common language for political scientists and AI/big data engineers to work together on common issues.

Academic compartmentalization and intellectual indifference discourage scholars with different background to collaborate. Before tackling complicated big data with AI technologies, we may have to remove the invisible barriers that separate the political scientists and the AI/big data engineers.