Recommended Selves: Authenticity in Algorithmic Filtering

By allocating their attention to pieces of content, algorithmic filtering shapes the daily behavior of billions of users when they interact with a digital platform. Beyond conditioning what we do, can recommendation algorithms influence who we are? This paper suggests that they do. Specifically, I contend that recommender systems affect users’ capacity to be their authentic selves in both positive and negative ways. I start by offering an account of authenticity that builds on two central concepts: volitional alignment and self-understanding. I then explain how algorithmic filtering works and impacts authenticity. While recommender systems frustrate users’ second-order desires by relying on uninformative behavioral signals, they also facilitate self-understanding by inciting users to question their identity. I end by discussing how controllable and explainable recommenders would best enable users to be authentic.