Understanding Responsible AI Stewardship in the Public Sector

This project was conducted as a collaboration between the Blavatnik School of Government, University of Oxford and the University of Sheffield to investigate the skills and trainings necessary for responsible AI stewardship in the public sector, using case studies from three OECD member countries.

The objectives of the research were to:

• Objective 1: Identify AI skills required by recent regulations and policy developments. This involved analysing regulatory and policy documents released in the UK and abroad over the last 12 months to pinpoint the AI skills and competencies necessary to comply with these requirements.

• Objective 2: Conduct interviews with twelve stakeholders in public, private sector and non-profit organisations, specifically in local government and research institutions/labs in UK, Canada and the Netherlands to learn about AI guidelines and skills, effects of regulations on startups and the availability or necessity for appropriate training provision.

The report's conclusions are based on a combination of academic research, consultations, and formal interviews conducted with various stakeholders over a three- month period. These insights were analysed within the context of the current regulatory landscape, with a focus on UK, Dutch, and Canadian governments, as well as international reports and recommendations from global organisations.

The research in this report was ethically approved under the University of Sheffield's Ethics Review procedure, as administered by the Information School (research ethics protocol no. 062396).