The (Non) Deus-Ex Machina: A Realistic Assessment of Machine Learning for Countering Domestic Terrorism

In light of the January 6 insurrection, the Department of Homeland Security (DHS) and other national security agencies are looking toward using more artificial intelligence (AI) and machine learning (ML) tools to detect and combat extremism in America. AI and ML hold much promise for the domestic CT mission, but the discourse has placed on them unrealistic expectations that do not conform to what is technically possible. This essay seeks to create a baseline conversation about what is ML, how it actually works, and what is a more realistic use case for ML in domestic CT. The core argument is that current ML tools are not optimal for the CT enterprise because terrorism experts are often sidelined in the development and the implementation of these algorithms.