Abstract

In April 2013, the American journalist Matt Taibi published an article in «Rolling Stones» entitled: Everything is rigged: the biggest price-fixing scandal ever about finance and conspiracy theories. In 2016, Hbo launched the dystopian series Westworld with the claim: «The world is a rigged game, so you best learn to play dirty» echoing sentiments and opinions circulated on social networks. The three main dimensions of this discourse are: the lack of rules and «ethics» in the political arena; the dimension of the game as a metaphor and as a narrative form of political action; the importance of new technologies and «machines» as an aid in this rigged game. The rise of Artificial intelligence (Ai) has taken this dynamic to even more sophisticated levels than before. In recent years, one of the most worrying challenges has been the impact of Ai-powered disinformation on public opinion, electoral processes and democratic accountability. While public attention and expert analysis has mainly focused on the generation of fake news or distorted narratives, the use of Ai has dramatically increased the scope and potentially the effectiveness of the individual voter behaviour manipulation and microtargeting techniques that political campaigns have used since the early 2000s (e.g., reinforcement learning, services like Clogger Inc.). This article aims to shed light on these different and complementary strategies of using Ai in the construction of electoral consensus with a specific focus on their role in the last general elections held in Spain on 23 July 2023. The political dynamics of the elections, with Pedro Sanchez’s government defending its actions against the right-wing opposition parties Partido popular and Vox, offer a relevant scenario to explore how Ai-generated content and tools can influence and shape electoral behaviour. The analysis will not only be limited to the identification of themes, narratives and patterns of Ai-generated disinformation, but will try to provide examples of what can be defined as the Ai-generated paradoxical reinforcement process.