Federal Procurement of Artificial Intelligence: Perils and Possibilities

In Federal Procurement of Artificial Intelligence: Perils and Possibilities, David S. Rubenstein proposes reforms to federal procurement law to establish transparent and ethical AI practices across federal agencies.

The proliferation of Artificial Intelligence (AI) use by federal agencies raises urgent questions about how these new technologies should be regulated. Today, AI procurement helps streamline processes in agencies like the Social Security Administration, the Food and Drug Administration, Homeland Security, and more. Under the right conditions, algorithmic governance can be an incredibly useful tool, however, under the wrong conditions, it can lead to widespread discrimination, invasion of privacy, and the degradation of democratic principles. Yet, much, if not most, of the AI used by federal agencies will be procured from a virtually unregulated private market. In this report, David. S Rubenstein shows that when the government acquires AI, it is often procuring the policy choices of the nongovernmental actors who designed the technology.

Rubenstein outlines a plan for ethical AI procurement, going forward: mandating the creation of a government-wide inventory that includes clear information on AI systems used by federal agencies, requiring agencies to prepare “AI risk assessment” reports prior to acquiring AI services, and integrating ethical AI consideration into existing regulations for source selection. Federal procurement of AI services must be reimagined as more than just a marketplace but rather a policymaking space that promotes trustworthy and ethical AI.