The Public Interest Declassification Board (PIDB) has noted that the executive branch has been developing and implementing artificial intelligence (AI) and machine learning (ML) technologies for a variety of use cases at an accelerated rate. However, federal agencies too often develop these innovations without internal or interagency coordination. Recommendations of the PIDB have long emphasized the urgency of coordinating AI/ML solutions and integrating them with the modernization of classification and declassification processes across the executive branch.
Created in December 2021, the new Department of Defense (DoD) position of Chief Digital and Artificial Intelligence Officer (CDAO), will advance the coordination of AI development within the DoD. This new position highlights the role of emerging technologies in accessing data for decision making and realigns DoD work on artificial intelligence and data management. Instead of relying on reported data, senior DoD leaders use real-time data and analytics to make decisions for personnel and readiness. A key component for DoD operations is data management. Without this management, data cannot be effectively deployed into workflows and missions to inform decision making. To leverage these capabilities, the new CDAO will report directly to the Deputy Secretary of Defense and oversee the Joint Artificial Intelligence Center (JAIC), the Defense Digital Service (DDS), and the DoD Chief Data Officer.
The JAIC has also just announced a new position for “AI Assurance,” to oversee artificial intelligence (AI) ethics policy, testing, and evaluation. The goal of this position is to ensure confidence in the algorithms now employed in over 600 AI projects and applications that range from detecting patterns in COVID-19 data to identifying cyber vulnerabilities.
The PIDB recommends developing AI use cases for the following potential applications to:
- Automate systems that would facilitate the sharing of classified information with allies, including the North Atlantic Treaty Organization;
- Support systems that make classification and declassification more precise and that enables consistent classification across components and agencies; and
- Permit systems to facilitate appropriate information sharing, including to troops in the field.
The PIDB recognizes the development of DoD interest in AI technologies as central to information security, classification management, and declassification requirements. Future projects and use cases for AI/ML could be used for managing and organizing classified records and data. Through automation, AI/ML could streamline the categorization of large volumes of data for integration into human decision making. Potential application development should include tools to avoid the over-classification of national security information, while supporting human reviewers in the declassification process, and managing information sharing among stakeholders.
PIDB 2020 Report to the President: https://www.archives.gov/files/declassification/pidb/recommendations/pidb-vision-for-digital-age-may-2020.pdf
Deputy Secretary of Defense Kathleen Hicks Memorandum to DoD Components: https://media.defense.gov/2021/Dec/08/2002906075/-1/-1/1/MEMORANDUM-ON-ESTABLISHMENT-OF-THE-CHIEF-DIGITAL-AND-ARTIFICIAL-INTELLIGENCE-OFFICER.PDF
Secretary of Defense Austin Lloyd’s speech: https://www.defense.gov/News/Speeches/Speech/Article/2695046/secretary-of-defense-austin-remarks-at-the-global-emerging-technology-summit-of/
Fedscoop article: https://www.fedscoop.com/jaic-creates-new-position-for-ai-assurance/