The problem of digital shadows in Plato’s cave

Using Plato’s allegory of the cave, this essay explores the problem of “digital shadows” by drawing parallels between its prisoners’ perception of reality and the way modern AI systems interpret human behavior. Just as the cave dwellers mistake shadows for reality, machine learning (ML) algorithms operate solely on digital traces—tokenized, reductive representations of human life. These shadows, while useful for pattern recognition, fail to capture the embodied, social, and historical dimensions of human existence. The use of AI (ML) in this manner risks constructing a world optimized for shadows rather than people—an improvised reality that, like Plato’s cave, blinds us to the truth of human experience. Examples from credit scoring and predictive policing are employed to illustrate how digital shadows can entrench inequality.