Election campaigns and normative goods in an AI-accelerated environment

Submitted for Rachel Gibson's Panel: This paper focuses on some of the normative democratic risks that may be at stake with the use of AI in election campaigns. Specifically, it considers the potential for AI-powered communications technology to become a necessity for campaigners to respond to continuously evolving messages by a) campaign opponents and b) sources of misinformation. The paper presupposes a relational view of election campaigning, where the core normative goal of election campaigns is for candidates and parties to build trust with the public based on their sincerity, values and competence. At least in part then, campaigns must be able to provide the discursive space in which these trust relations can be built. New technologies in the past have typically changed this discursive space – at one time favouring those best able to design glossy pamphlets and at others those best able to perform on television. Rarely, however, has technology threatened to erode that space. AI-powered communications technology threatens to do this in three ways, namely by a) saturating discursive space with AI generated content (whether accurate or not) b) making it more difficult to identify content endorsed by a party or candidate and c) making it less clear when a candidate themselves is truly speaking, truly tweeting, or truly saying what they believe. Intriguingly, AI puts a new twist on existing problems with trust-building in the campaign context, where the autonomy and sincerity of candidates and parties are often called into question to the extent that they become driven by polling data and electoral strategy over and above political commitment. In an AI-accelerated environment, the autonomy and sincerity of candidates and parties becomes all more difficult to perceive.