Responsible Adoption of Artificial Intelligence in Institutional Review Board Proposal Review. A Narrative Review

Abstract

Background: Institutional Review Boards (IRBs) are central to protecting human research participants, yet they contend with inconsistent decision-making, protocol-approval delays, and resource constraints that strain the quality and timeliness of ethics oversight. Artificial intelligence (AI) — particularly application-specific large language models (LLMs) and machine-learning tools — has been proposed to ease these pressures while preserving ethical rigour. Objective: To synthesise current evidence and set out a practical, responsibly governed framework for integrating AI into IRB proposal review in a large tertiary-care setting. Methods: We conducted a narrative review of 15 peer-reviewed publications (2015–2026) — empirical studies, conceptual frameworks, and expert commentary on AI in research ethics — and mapped the evidence onto the IRB review workflow, organising it around implementation, technical architecture, governance, risk, and evaluation. Results: The evidence indicates that AI can shorten specific review tasks (an approximately 25% reduction in task time) and support consistency, achieving high accuracy on well-defined tasks (for example, recall 0.90 and precision 0.87, with 91% of machine-generated risk-of-bias judgements accepted) when human oversight is retained. The highest-value, lowest-risk entry points are completeness screening, protocol summarisation, and triage; consistency checking, decision support, and consent-related applications demand stricter validation. Realising these benefits depends on human-in-the-loop safeguards, clear accountability and transparency, mitigation of bias and automation bias, privacy protection, and regulatory compliance. Conclusion: Used to augment rather than replace human ethical judgement, AI can make IRB review faster, more consistent, and more thorough. A phased, pilot-driven pathway with predefined success criteria — attentive to small institutions and to low- and middle-income settings — offers a responsible route to adoption.