Exploring the operational challenges of navigating ethical oversight in the era of artificial intelligence: a qualitative study of health research ethics committees in Tanzania

Abstract

BACKGROUND: Health Research Ethics Committees (HRECs) play a pivotal role in safeguarding research participants and ensuring ethical conduct. The rapid integration of artificial intelligence (AI) into health research introduces novel ethical and operational complexities, including algorithmic opacity, bias, data governance challenges, and difficulties in post-approval monitoring. However, empirical evidence on how these AI-specific complexities affect HREC operations in Tanzania remains limited. This study explored operational challenges associated with ethical oversight of AI-related health research among HRECs in Tanzania. METHODS: An exploratory qualitative study design was employed, involving 25 participants (15 HREC members and 10 secretariat staff) purposively selected from 10 HRECs across five zones of mainland Tanzania. In-depth interviews were conducted using a semi-structured guide. Data were transcribed, translated, and analyzed using inductive content analysis with thematic interpretation, supported by NVivo software. An audit trail, reflexive journaling, and data source triangulation were used to enhance credibility and trustworthiness. RESULTS: A total of 28 codes were generated and organized into 10 subthemes and four overarching themes. While general challenges such as limited funding, high workload, and staffing constraints persisted, AI-related protocols introduced additional operational burdens, including difficulties in assessing algorithmic validity, increased reliance on external technical experts, and challenges in reviewing large-scale datasets. Although 50% of secretariat staff had more than five years of experience, participants emphasized that the key limitation was not general experience but insufficient AI-specific technical expertise. Weak post-approval monitoring systems were particularly inadequate for tracking AI-driven studies. CONCLUSION: Tanzanian HRECs demonstrate foundational governance capacity but face AI-specific operational and technical challenges that constrain effective oversight. Strengthening AI-focused training, technical advisory mechanisms, digital review systems, and sustainable financing is essential to support the ethical governance of emerging technologies.