Examining stakeholders’ involvement in AI policy initiatives: The case of the United Kingdome

Artificial Intelligence (AI) algorithms have become a prominent tool to govern many aspects of our public life. Despite its efficiency and the promise of enhanced public policies, the use of AI is not neutral. Most governments are developing AI strategies and initiatives stressing the need to meet specific principles in order to be deemed trustworthy. One key element often mentioned in these initiatives is the need to ensure diversity, non-discrimination, and fairness, which calls for the involvement of citizens and stakeholders through the entire life cycle of AI tools. However, research and practice have very limited knowledge on how citizens and stakeholders are involved in AI policy initiatives. To address this gap, the present paper conducts a thematic content analysis of 30 AI policy initiatives produced by (semi-)public organizations in the UK in the last 5 years. The analysis focuses on three questions related to the engagement of stakeholders and citizens in policy initiatives: (1) what is the goal of stakeholder and citizen engagement? (2) who is invited to participate in the process? (3) how is the participation structured and implemented in practice? Our preliminary results indicate that, whereas AI policy initiatives reflect on the need to include different type of stakeholders to promote certain values such as transparency, fairness, inclusiveness and equity, they rarely specify the stakeholders and citizens that should take part in this process. Furthermore, our thematic content analysis conducted using Atlasti software showed that very few documents specify how this participation should take place. Overall, our research demonstrates that, despite the rhetoric that calls for more open and transparent AI initiatives, we still see very limited progress in practice.