Evidence-Based Policy, Artificial Intelligence, and the Ethical Practice of Policy Analysis

Dr Joshua Newman and Professor Michael Mintrom introduce the second article – Mapping the Discourse on Evidence-Based Policy, Artificial Intelligence, and the Ethical Practice of Policy Analysis  – to be published in the Journal of European Public Policy Special Issue ‘The Politics of Policy Analysis’. They explore the role of artificial intelligence (AI) as a new technology that may encourage old ideas about policy analysis. The ability to use AI, in tandem with ‘big data’, to process huge amounts of policy relevant information, raises (again) the prospect that key parts of decision-making can be routinised and removed from politics. Yet, applications so far show that each aspect of that process contains – or hides – a multitude of political decisions that should be surfaced to allow proper debate and routine accountability.

Evidence-based policy is a hotly debated topic. Supporters argue that public sector decision making is in bad shape, influenced primarily by ideological thinking, pressure from special interest groups, and heavy demands on resource-poor public servants who are frequently asked to provide crucial advice within short timeframes. Critics argue that information is subjective, and decision-making is necessarily political, so evidence-based policy is in any case both unachievable and undesirable. However, this is where the debate has stalled.

We are rapidly entering an age of advanced computer systems that can recognise patterns, analyse large datasets, and autonomously improve their own programming, functions that are often referred to as ‘artificial intelligence’, or AI. The use of AI in the public sector is on the rise, in areas of service delivery as diverse as education, traffic management, and criminal justice.

What impact will AI have on how we think about evidence-based policy? Can we call information generated by computer algorithms, ‘evidence’? Are we prepared to deal with the ethical concerns inherent in letting computers inform decisions with material consequences for the lives of ordinary citizens and service users?

In this article, we argue that in light of advances in AI, debates about evidence-based policy will need to be updated. By looking at different arguments in support of and critical of evidence-based policy, and the various concerns that have been raised with respect to the ethical dilemmas related to using AI for public service delivery, we outline eight different directions in which the debate could advance. Then, using the SyRI welfare fraud detection scandal that brought down the government in the Netherlands in 2021 as an illustrative example, we show how different perspectives on evidence can actually be combined in a way that lets us see many sides of a complex issue at once. Discussions about the use of — or even the existence of — evidence in public sector decision making may already be lively, but the advent of AI threatens to make these debates even more competitive. However, it is possible that arguments that seem to be at odds could be made to work together, to support a more holistic understanding of how computers and automation can influence decision making, and how to prepare for policy controversies in an AI-enabled future.

Newman, J. and Mintrom, M. (2023) ‘Mapping the Discourse on Evidence-Based Policy, Artificial Intelligence, and the Ethical Practice of Policy Analysis’, Journal of European Public Policy, https://www.tandfonline.com/doi/full/10.1080/13501763.2023.2193223