Domenique Lumpkin

Key Findings Datasets and knowledge representations need to be carefully curated. High-quality datasets must be built with care and must also be associated with the right metadata to support object-based production and subsequent algorithm development. Analysts can and should anticipate AI failure modes. Understanding the limits of an AI system’s training data or knowledge representation will help analysts anticipate errors and use the AI systems more effectively.

This research was sponsored by the Defense Intelligence Agency and conducted within the International Security and Defense Policy Program of the RAND National Security Research Division. This publication is part of the RAND research report series. Research reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND research reports undergo rigorous peer review to ensure high standards for research quality and objectivity.