Epistemic Injustice in and through AI

Abstract

AI is rapidly transforming knowledge production and practices across a range of domains, yet AI technologies often embed and perpetuate epistemic injustices—privileging dominant perspectives while marginalising others. Despite growing awareness of AI biases, many frameworks used in HCI and AI ethics fail to fully account for how AI models reproduce historical and systemic exclusions. We propose to critically examine epistemic injustice in AI across six domains; generative AI, creative practice, healthcare, work, education and automated decision-making. We explore how AI systems respond to diverse sociocultural, linguistic, and epistemological inputs, revealing biases in representation, accessibility, and credibility. Through reflection and collaborative mapping, we aim to identify research priorities and intervention strategies at individual, community, and broader society levels. By fostering rich dialogue and nuanced evidencing, we seek to advance research on epistemic justice in AI and create pathways for more inclusive and equitable futures.