Participatory Assessment of Ethics in AI: A Review

Artificial Intelligence (AI) is reshaping digital healthcare by advancing disease detection, treatment options, and individualized patient care. In this context, responsible AI emphasizes building systems that follow ethical standards, reduce bias, and safeguard personal data, preserving patient safety and dignity. Moreover, its algorithms should be clear and open to evaluation. Although it brings major opportunities, responsible AI still encounters notable obstacles and limitations.

To understand the nature of participatory AI ethics assessment, from the nature of the evaluators to the way in which ethics are assessed.

A scoping review of literature on evaluation of responsible AI was performed using several scientific databases.

7 papers were found that explicitly mentioned "ethics" in their titles or abstracts and reported on studies evaluating responsible AI systems using a participatory approach. The studies show considerable heterogeneity in terms of the experts who participated in the assessment. Only one study involved patients. The factors evaluated also show a high degree of heterogeneity. The most used participatory methods were interviews and surveys.

participatory studies assessing ethics in AI should report information in detail about participants. Participant background, gender, and level of expertise should be reported, together with the multidisciplinary character or not of the study, and the methods used to collect information. Furthermore, ethics assessment of AI should consider ethical principles contained in common frameworks and should use an unambiguous nomenclature.