Public Procurement of AI - Enabling Innovation and Managing Risks?

Session overview

As AI becomes a cornerstone of digital transformation in government, public sector organisations face a new frontier of procurement complexity. This session will explore how public procurement can support, or hinder, the responsible acquisition of AI technologies. We’ll examine the tensions between fostering innovation and ensuring accountability, the unique challenges of safety, security, and risk in AI systems, and what meaningful transparency and KPIs might look like in practice.

This is a significant challenge and successful AI procurement will require a new set of enhanced capabilities and approaches to market interactions. The UK Government has been working on centralised programmes, such as 'Test and Learn', an incubator for AI (i.AI), or a newly-announced National Digital Exchange (NDX). Local government organisations, such as the Local Government Association, have issued guidelines and are fostering knowledge-exchange initiatives. At the moment, it is unclear what this will mean for AI procurement in the longer term.

Should AI procurement be centralised for efficiency, or decentralised for context? And what does ‘responsible AI’ mean when outcomes are uncertain and evolving? How should we collectively develop helpful knowledge repositories and how could we best ensure dissemination of emerging best practices while learning from the unavoidable mistakes and failures in such a fast-paced developing area?

Our excellent panellists are: 

• Prof Albert Sanchez Graells, Professor of Economic Law, University of Bristol Law School
• Dr Cari Miller, Head of AI Governance and Research, The Center for Inclusive Change
• Dr Gisele Waters, Co-Founder, AI Procurement Lab
• Kaye Sklar, Senior Program Manager for Content and Insights, Open Contracting Partnership
• Dr Mavis Machirori, Senior Researcher, Ada Lovelace Institute
• Rachel Coldicutt OBE, Founder and Executive Director, Careful Industries
• Saema Jaffer, Head of Commercial Policy and Capability, UK Parliament

This online session is co-hosted by the University of Oxford’s Faculty of Law and the Government Outcomes Lab (GO Lab) in the Blavatnik School of Government.

Resources

• AI Playbook for the UK Government:
  A foundational guide outlining the UK Government’s approach to adopting and governing AI technologies. Read the AI Playbook here.
• Albert Sanchez-Graells – Responsibly Buying Artificial Intelligence: A ‘Regulatory Hallucination':
  This open-access article critiques the assumption that public buyers can confidently and responsibly procure AI technologies under current conditions, warning of both individual harms and systemic risks. Albert proposes an alternative strategy involving the creation of an independent authority. Read the article here.
• Consequence Scanner for AI Procurement:
  A collaborative project between Albert Sanchez-Graells and Rachel Coldicutt, focused on developing a tool to anticipate and assess potential consequences in public sector AI procurement. Find out more here.
• Felix-Anselm van Lier – Government AI Adoption: Formal-Relational Contracting Can Unlock Responsible Progress:
  In this blog post, Felix-Anselm van Lier (Research and Policy Fellow at the Government Outcomes Lab) explores how a flexible, collaborative contracting model can help governments adopt AI responsibly, balancing innovation with accountability. Read the blog here.