Acquiring Ethical AI

73 Pages Posted: 18 Nov 2020 Last revised: 7 Dec 2021

Date Written: November 1, 2021


Artificial intelligence (AI) is transforming how the federal government operates. Under the right conditions, AI systems can solve complex problems, reduce administrative burdens, improve human decisions, and optimize resources. Under the wrong conditions, AI systems can lead to widespread discrimination, invasions of privacy, and the erosion of democratic norms. A burgeoning literature has emerged to square algorithmic governance with the precepts of constitutional and administrative law. Federal procurement law, however, remains a dangerous blind spot in the reformist agenda. This Article pivots into that neglected space and emerges with comprehensive framework for acquiring ethical AI. Toward that end, the Article makes three main contributions. First, it provides an original account that yokes the ambitions of algorithmic governance, the imperative of ethical AI, and the levers of procurement law. Second, this Article argues that the procurement system is uniquely situated to check and enable algorithmic governance in ways that other legal frameworks miss. Third, the Article prescribes a set of concrete regulatory reforms to instantiate ethical AI throughout the procurement process: from acquisition planning to market solicitation, bid evaluation, source selection, and contract performance. Procurement law will not solve all the challenges of algorithmic governance. Just as surely, those challenges cannot be solved without procurement law.

Keywords: artificial intelligence, machine learning, AI ethics, procurement, algorithmic governance, administrative law

Suggested Citation

Rubenstein, David S., Acquiring Ethical AI (November 1, 2021). Florida Law Review, Vol. 73, 2021, Available at SSRN:

David S. Rubenstein (Contact Author)

Washburn University - School of Law ( email )

1700 College Avenue
Topeka, KS 66621
United States
785-670-1682 (Phone)

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
PlumX Metrics