Electoral administration and state-of-the-art AI: a cocktail of incongruences

Artificial Intelligence (AI) is rapidly spreading into various avenues of human activity. This article considers the suitability of state-of-the-art AI for use within the sector of electoral administration, which comprises processes critical to the sustainment of healthy democratic institutions and procedures We posit that there are fundamental incongruences between the ways of current AI and the values and principles that electoral administration processes need to adhere to. This article explores the contours of such incongruences using specific examples of (potential) risks from AI usage within electoral administration. We outline how state-of-the-art AI has an entrenched majoritarian bias, data-first ethos, neoliberal propensity and opaque operation, all of which pose significant hurdles for usage within electoral administration. While our intent is primarily limited to conveying the nature of such incongruences to a non-technical audience, we also briefly deliberate on possible pathways towards mitigating such harm.