Machine Learning (ML) models are increasingly applied to wearable self-tracking technologies to offer daily classifications and recommendations for well-being. This shift introduces design challenges, particularly regarding the opacity of training processes and model outputs. We contribute to this space with a conceptual framing of the algorithmic gaze on body and well-being, which we use to critically investigate long-term engagement with a wearable self-tracking technology. Through an autoethnographic study with the Oura Ring, we identified three themes, highlighting tensions between wearer and the ML models, namely: Conflicting narratives of daily activities, fine-tuning of the human, and blurry boundaries of multiple bodies using such devices simultaneously. Departing from the themes, we used fabulation as a method to craft narratives that probe the tensions from the algorithmic gaze, from which we offer alternative design openings for ML in wearable self-tracking devices.
ACM CHI Conference on Human Factors in Computing Systems