Artificial intelligence and the promise of participant-centered free and informed consent

Abstract

Informed consent remains a persistent challenge in clinical research, as traditional processes often fail to ensure participant understanding and voluntariness. We developed and evaluated an artificial intelligence (AI)-driven consent process utilizing large language models to conduct interactive, participant-centered consent conversations. The system’s scalability, auditability, and capacity to tailor communication may address barriers to informed consent. In silico testing with simulated participants demonstrated high rates of successful teach-back across consent concepts, with auditability and adherence to predefined ethical and procedural safeguards. These simulation-based results support pilot testing of AI-driven consent processes to evaluate their potential to enhance participant understanding, autonomy, and trial accessibility.