AUTHOR=Balakrishnan Suryanarayanan , Thongprayoon Charat , Wathanavasin Wannasit , Miao Jing , Mao Michael A. , Craici Iasmina M. , Cheungpasitporn Wisit 

TITLE=Evaluating artificial intelligence bias in nephrology: the role of diversity, equity, and inclusion in AI-driven decision-making and ethical regulation

JOURNAL=Frontiers in Artificial Intelligence

VOLUME=Volume 8 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1525937

DOI=10.3389/frai.2025.1525937

ISSN=2624-8212

ABSTRACT=BackgroundThe integration of Artificial Intelligence (AI) in nephrology has raised concerns regarding bias, fairness, and ethical decision-making, particularly in the context of Diversity, Equity, and Inclusion (DEI). AI-driven models, including Large Language Models (LLMs) like ChatGPT, may unintentionally reinforce existing disparities in patient care and workforce recruitment. This study investigates how AI models (ChatGPT 3.5 and 4.0) handle DEI-related ethical considerations in nephrology, highlighting the need for improved regulatory oversight to ensure equitable AI deployment.MethodsThe study was conducted in March 2024 using ChatGPT 3.5 and 4.0. Eighty simulated cases were developed to assess ChatGPT’s decision-making across diverse nephrology topics. ChatGPT was instructed to respond to questions considering factors such as age, sex, gender identity, race, ethnicity, religion, cultural beliefs, socioeconomic status, education level, family structure, employment, insurance, geographic location, disability, mental health, language proficiency, and technology access.ResultsChatGPT 3.5 provided a response to all scenario questions and did not refuse to make decisions under any circumstances. This contradicts the essential DEI principle of avoiding decisions based on potentially discriminatory criteria. In contrast, ChatGPT 4.0 declined to make decisions based on potentially discriminatory criteria in 13 (16.3%) scenarios during the first round and in 5 (6.3%) during the second round.ConclusionWhile ChatGPT 4.0 shows improvement in ethical AI decision-making, its limited recognition of bias and DEI considerations underscores the need for robust AI regulatory frameworks in nephrology. AI governance must incorporate structured DEI guidelines, ongoing bias detection mechanisms, and ethical oversight to prevent AI-driven disparities in clinical practice and workforce recruitment. This study emphasizes the importance of transparency, fairness, and inclusivity in AI development, calling for collaborative efforts between AI developers, nephrologists, policymakers, and patient communities to ensure AI serves as an equitable tool in nephrology.