Algorithmic Bias in AI-Driven News Production and Dissemination: The Dynamics of X Misinformation During Elections

Abstract

Society uses information from different sources for different reasons, including decision-making and civic engagement. With the information explosion in the early 1950s, libraries and other information systems are grappling with what constitutes relevant information to help users’ effective information retrieval (IR). The situation grew worse with social media's advent in the early 2000s. With media now in the mix of the information chain, the user behavior and the role of librarians have been redefined. Political participation by citizens through information provided on social media, particularly X’s platform, has gained traction for election discussions and information sharing in recent times. However, the AI-driven X’s recommendation algorithms, particularly the SimClusters segment, perpetuate misinformation, create and reinforce polarizations, and consequently, threat to world democracies. The SimClusters detect communities through a bipartite graph with a set of nodes representing users and interests using content similarities. It generates communities based on the weight of the nodes' interests through interactions such as views, retweets, likes, comments, and follows. In the process of addressing questions like what tweets and users are similar to my interests, and what tweets my friend liked, confirmation bias results, and echo chambers that create polarization during election discourse. The study surveyed primary and secondary sources from JSTOR and ProQuest databases on X’s election information sharing, which revealed that X’s platform perpetuated misinformation, amplified subjective and divisive narratives during the elections in the US, EU, Germany, and Poland. Reimagining the recommendation algorithms to incorporate a fact-checking component would address misinformation, confirmation bias, and echo chambers, and unify information sharing. This would reshape our democratic engagement and build an enduring future through equitable access to information.