Abstract
Researchers and policymakers call for the development of ethical AI. Within Information Science, significant attention has been given to the responsible use and governance of AI. However, less focus has been given to building ethics into AI in the sense of making AI systems to act morally responsibly without human intervention. This cross-disciplinary systematic review examines empirical studies published in scientific journals between 2016 and 2024 to explore what characterizes the efforts of building ethics into AI. Using a comprehensive systematic review method, 6,752 articles were identified, of which 59 were selected and analysed by a multidisciplinary research team. Findings show that the practical development of ethical AI remains largely experimental. Most studies adopt a “thin” conception of ethics, focused on technical operationalizations of a single moral principle or an aspect of ethical theory. This limited ethical scope may reflect the predominance of computer scientists in the field and a lack of sustained, iterative collaboration with ethicists and social scientists. The study concludes that if ethical AI is possible and justified, it will require a shift toward a more pluralistic and interdisciplinary approach to ethics.
Recommended Citation
Flatås, Bjørn A.; Nordsteien, Anita; Eide, Hilde; Lysdahl, Kristin B.; Sanchez, Veralia G.; Stendal, Karen; Turk, Eva; and Eide, Tom, "Building Ethics into Artificial Intelligence: A cross-disciplinary systematic review" (2025). SJIS Preprints (Forthcoming). 12.
https://aisel.aisnet.org/sjis_preprints/12