Computational psychiatry (CP), based on artificial intelligence technology, plays an increasingly important role in scientific research and medical diagnosis. Epistemic concerns in the ethics of artificial intelligence have also been at the center of debate in CP, but the different epistemic forms of injustice caused by the internal cultures of CP remain unexplained. We distinguish between data-driven and theory-driven cultures and their research purposes via practical examples of CP models deployed in addiction. A data-driven culture may advance medical understanding of biological categories of mental illness, whereas a theory-driven culture provides better explanatory mechanisms between symptoms and biology. We discuss testimonial injustice caused by the silencing of patient voices in a data-driven culture, and hermeneutic injustice caused by the non-sharing of hermeneutic resources in theory-driven culture based on Miranda Fricker's account of epistemic injustice. We analyze the factors underlying nuances in epistemic forms between the two, such as naturalistic-dominated medical understanding and the system's epistemic privileging. The above epistemic risks all indicate the intricacies of mental disorders and require that success be assessed in terms of actual benefit to patients. Finally, we emphasize the importance of the patient's phenomenology and call for greater inclusion of patients in psychiatric decision-making processes.
Wang, M. ., Wu, Z. ., Huang, L. ., Zhang, X. ., & He, X. . (2024). Computing addiction: Epistemic injustice challenges in the culture of computational psychiatry. Acta Bioethica, 30(2), 263–272. Retrieved from https://lajtp.uchile.cl/index.php/AB/article/view/76142
