Killoran, JayPark, Andrew2026-06-252026-06-252026Killoran, J., & Park, A. (2026). Learn from the blame game when AI causes harm. Proceedings of the National Academy of Sciences - PNAS, 123(17), Article e2528408123. https://doi.org/10.1073/pnas.2528408123https://doi.org/10.1073/pnas.2528408123https://hdl.handle.net/1828/24011In 2023, families of deceased Medicare Advantage patients filed a class action lawsuit against UnitedHealth Group, alleging that an AI algorithm, nH Predict, systematically denied medically necessary post-acute care. Physicians had approved the care, but nH Predict denied it. The lawsuit alleged that UnitedHealth knew the tool had a 90% error rate, yet continued to deploy it, counting on fewer than 1% of patients appealing. For its part, UnitedHealth has denied wrongdoing, maintaining that nH Predict was never used to make coverage determinations but, rather, served only as a clinical guide and that all coverage decisions were made by human reviewers. The case, which is ongoing, ignited a wider debate: Who was to blame? The engineers? The company? The case managers? Or the AI system?enAttribution-NonCommercial-NoDerivatives 4.0 InternationalArticlePeter B. Gustavson School of Business