Artificial intelligence (AI) models prioritize starkly different attributes than humans when making high-stakes decisions, and they don't express indecision like humans do, according to a new study led by Penn State researchers, raising questions about the role of AI in decision support for ethically sensitive situations like medical decisions.