Managing elections in the AI age: capacity, context and control

Although AI is increasingly embedded across almost every area of society, its application for electoral management bodies (EMBs) remains understudied. In this paper we address that gap using the oversight of political financing as a case study within the wider functions of EMBs. Drawing on ten semi-structured interviews alongside a sustained engagement with EMBs and civil society organisations through workshops and working groups over the past three years, we outline how EMBs are using AI tools, but also the wider barriers to their adoption. Whilst across cases AI has a recognised potential to improve transparency, enhance data granularity and reduce administrative burdens for both regulators and regulated communities, concerns still remain. These are largely centred around trust, both within EMBs themselves and among wider society, as well as wider issues centred around digital literacy and resourcing. We conclude by proposing a system of best practice to manage these problems which emphasises context-sensitive design practices and embedding a “human in the loop” philosophy at every stage of adoption to safeguard democratic accountability.