CSET provides policymakers with nonpartisan analysis of the security impacts of emerging technologies. Our initial research agenda focuses on artificial intelligence and advanced computing.

Our research is informed by the work of our Data Science Team, which gathers, processes, and interprets data at a larger scale than previous open-source assessments, including foreign-language and technical materials from diverse sources.

Themes and questions that we plan to address include:

  • National competitiveness. Comparisons of AI capabilities across countries typically rely on a patchwork of indicators, including access to data, investment levels, publication counts, and hardware production. What combination of these metrics best shows how countries compare across different areas of AI development? How do more comprehensive data on each metric change our understanding? How should commercial success be viewed in comparison to fundamental research progress, and how do each of these contribute to economic or military power?

  • Talent and knowledge flows. Innovation in AI and other emerging technologies relies on open research and mobile talent. In previous eras, governments were able to limit access to consequential research results, restricting them to government-sponsored labs and contractors. Today, the free flow of ideas, research results, software frameworks, and human capital knows no national boundaries. What should trade and immigration policies look like in this context? How can companies, universities, and governments best protect information from theft and misuse? How should governments manage the tension between their preference to control the best research, and researchers’ preference to publish openly?

  • Relationships with other technologies. Advances in AI and advanced computing will affect and be affected by other technologies, including many that are central to economic and military competition. What lessons can be drawn from past experiences managing the interaction of newer technologies (like cybersecurity) with older technologies (like nuclear weapons)? How will AI affect these technologies in the future? How can nations reduce the risks of unintended interactions between AI and other systems, including those of adversaries?