Research Ethics Committees (RECs) have a legal remit to make ethical judgements through their reviews and decisions on human subjects research. Yet many –if not most– RECs lack ethical/moral expertise within their composition, calling into question the rigour and standard of their normative reflections and deliberations in their judgments, and ultimately, their capability to safeguard research participants effectively.
The limited depth and quality of their normative analysis, interpretation, and reasoning (if present at all) hinders RECs’ ability to accurately identify, engage, and weigh competing moral considerations, and thus challenges their mandate to obtain appropriate moral judgement(s) which are inescapable in the rendering of ethical judgements. Left unaddressed, this deficiency risks undermining the effectiveness of the research ethics oversight system and the protections it is meant to afford.
Generative Artificial Intelligence (GenAI) tools –such as reasoning models (e.g., OpenAI’s o-series) and ‘Deep Research’– can arguably match the knowledge and skills of human experts, if appropriately trained. Such tools could address the lack of moral expertise in REC deliberations, and importantly, respond to the lack of depth and quality in their normative reflections and deliberations.
In this work-in-progress, I argue –on grounds of RECs’ prevailing lack of moral expertise– RECs should consider deferring to the outputs of a well-constructed, appropriately trained REC-specific GenAI to render ethical judgements. Doing so could enable RECs to arrive at appropriate moral judgements that are robust (i.e., derived from relevant moral expertise) and ethically justifiable/defensible. I will conclude by addressing key potential objections to this account.