Artificial Artificial Intelligence: Measuring Influence of AI 'Assessments' on Moral Decision-Making

AI feedback can influence moral decisions even when it's random: The study found that participants' judgments about kidney allocation were influenced by feedback they believed came from AI, regardless of whether that feedback was actually based on any real analysis. This suggests people may defer to AI recommendations on serious ethical decisions without sufficient scrutiny.

People treat AI and human expert feedback differently: The research revealed distinct differences in how participants responded to identical feedback depending on whether it was labeled as coming from AI versus human experts, indicating that the source of information—not just its content—shapes decision-making in morally significant contexts.

The "AI halo effect" may undermine human judgment: By showing that random AI assessments can sway decisions about life-or-death medical allocations, the study highlights a concerning gap between people's trust in AI systems and those systems' actual reliability, particularly in high-stakes ethical scenarios.

How AI feedback is presented matters for responsible deployment: The findings underscore that the design choice of how and when to present AI assessments to users has real consequences for moral decision-making, suggesting companies and institutions need careful guidelines before deploying AI in advisory roles for consequential decisions.