Public Sector AI Governance: Discrimination Awareness in Swedish Public AI Development

Description

This paper explores the ADM-related work conducted by the Swedish Equality Ombudsman (DO) that is currently invested in contributing to ADM development from an anti-discrimination perspective. In their effort to develop an alternative, equality-based vision for ADM in the public sector, DO conducted a survey to map ADM work and fundamental principles of guiding the ADM implementation in Swedish national authorities.

The survey explored i) to what extent national Swedish authorities are developing or utilizing ADM or AI-systems, and if they do, ii) to what extent they are taking issues of discrimination into account when doing it. With this, we are interested in iii) how the principled notions of AI governance are related to regulation on anti-discrimination when it comes to public sector AI development. The survey was conducted in late 2021 and early 2022, including 34 national public authorities.

In parallel, there has been a strong international push for governance of AI in terms of ethics guidelines or other principled documents, often stating the need for ensuring “fairness” in AI-systems. Much of the public administration is however already firmly regulated and governed. In this article, we focus on discrimination as governed by the Swedish Equality Ombudsman.

Based on survey data and the written report, and exchanges with DO, we situate their work within the broader move to govern ADM implementations and explore the alternative visions for ADM that the Ombudsman develops. By way of doing so we also map out the politics and structures of ADM governance development that involves several actors struggling for visibility and power of problem definition as well as proposing solutions.

Period	2022 Aug 29
Event title	Reframing ADM: Concepts, Values, Alternatives
Event type	Conference
Location	Copenhagen, DenmarkShow on map
Degree of Recognition	International