Artificial intelligence model distillation (that is, the process of training a new model on the outputs of a more capable one) poses a growing challenge to efforts to preserve the U.S. frontier model advantage. In this paper, the authors examine whether technical counter-distillation measures are likely to be sufficient to deter distillation. They argue that technical measures alone (such as by materially raising the cost of distillation or substantially reducing the quality of the resulting models) are unlikely to change the underlying incentives that make distillation attractive and that broader policy tools might be needed.
This publication is part of the RAND expert insights series. The expert insights series presents perspectives on timely policy issues.
This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.
RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
