Can the robot procure for you? -- a short reflection a propos Chesterman (2021)

origin.jpg

I am reading the very interesting new book by Simon Chesterman, We, the Robots? Regulating Artificial Intelligence and the Limits of the Law (Cambridge University Press, 2021). One of the thought-provoking issues the book addresses is the non-delegability of inherently governmental functions to artificial intelligence (AI). And one of the regulatory analogies used in the book concerns the limits to outsourcing as regulated by procurement law (see pages 109 ff).

The book argues that, for ‘certain decisions, it is necessary to have a human “in-the-loop” actively participating in those decisions’ (109), and states that to reach such determination, a ‘useful analogy is limits on government outsourcing to third parties’ (110).

In that regard, Chesterman leads us to consider the US approach to establishing ‘inherently governmental’ functions for the purposes of outsourcing, on which there is a very detailed and useful Office of Federal Procurement Policy (OFPP) Policy Letter 11–01, Performance of Inherently Governmental and Critical Functions.

I was curious to see whether procurement itself was considered an inherently governmental function not susceptible of outsourcing, and was glad to find this very nuanced specific treatment of the issue (I can hear my US colleagues and friends laughing at my previous ignorance).

Inherently governmental.png

It seems thus arguable that, in the US context and unless the decisions of an AI are somehow re-attributed to a Federal Government employee by some legal fiction, some aspects of procurement decision-making (at contract formation phase, but not only) cannot (yet?) be delegated to an AI (or fully automated. let’s say).

Which prompts me to reflect on what would be the treatment under EU law—and under different Member States’ approaches to constraints based on public functions and the exercise of public powers. This may be the seed for a research paper — or perhaps just a follow-on blogpost — but I would be very interested in any thoughts or comments, particularly if this is an issue someone has already thought or published about! As always, feedback and engagement most welcome at a.sanchez-graells@bristol.ac.uk.