Can ChatGPT follow an algorithm for ethical decision-making in public health?

Abstract

Abstract Background With the rapid development of AI, especially chat bots based on Large Language Models like OpenAI ChatGPT and Google Bard, the question of the day is can AI replace (partially or entirely) experts in certain fields - for example in Medical Ethics. The aim of this work is to test the application of ChatGPT to public health ethics (PHE) cases. Methods ChatGPT was given the task to analyse Ebola quarantine case in Sierra Leone by applying methodology for ethical analysis of its own choice and by applying a 4-step model for ethical analysis developed in Medical University - Pleven. The tests were done with simple and detailed tasks. No additional plugins to ChatGPT were included. Results Given a simple task ChatGPT applied 4-step model to the case - Facts, Stakeholders, Ethical principles (Autonomy, Non-maleficence, Beneficence, Justice), Potential consequences. A brief argumentation was provided for each point. ChatGPT identified key ethical concepts for the case (limiting autonomy and freedom, limiting access to resources). When asked to analyse the case with the principles of PHE as defined in The Cambridge Textbook of Bioethics, ChatGPT still used Beauchamp and Childress principles of bioethics. When introduced to PHE principles ChatGPT identified limiting access to resources, creating social and economic hardship for those affected by quarantine as key issues. ChatGPT managed to apply the 4-step model for ethical analysis developed in Medical University - Pleven to some degree. It produced detailed discussion but didn't use any ethical standards, patient's rights or public health policies that could help with decision-making. Individual autonomy was disregarded by ChatGPT in the most detailed analysis. Conclusions ChatGPT has basic knowledge of methodologies for ethical case analysis. With proper guidance and detailed instructions, it could follow complex methodologies and produce a decision. Key messages • AI will continue to develop and its inclusion in medicine and public health practice and teaching is inevitable. • Public health professionals should always be critical of AI decisions and double check the information that was provided by AI.