Using ChatGPT in ethical dilemmas and policy-related complex decision making: Are we ready yet?

Background and Objective: Artificial intelligence (AI) algorithms using language models have emerged as valuable tools in medicine. While AI has demonstrated its ability to address clinical questions, its application in ethical dilemmas remains debated. Some argue that AI can synthesize diverse information to form a comprehensive perspective, while others caution against premature reliance. This study explored the potential of AI in addressing ethical medical dilemmas faced by physicians, transitioning from theoretical discussions to practical solutions. Methods: ChatGPT-3.5 was presented with three socio-ethical dilemmas relevant to national health policy decisions, and its responses were compared to those of physicians and real-world decisions. The dilemmas included questions on (1) criteria for allocation of technologies when resources are limited (2) personalized treatment, and (3) conflicts between patient requests and health organizations' strategy. Results: ChatGPT-3.5 aligned with physicians' views on budget allocation but diverged on age-related criteria. It struggled to resolve conflicts between patient preferences and organizational strategies. Its responses reflected physician paternalism and a private market perspective, emphasizing system-wide benefit (utilitarian approach), likely due to familiarity with private healthcare systems. Conclusions: ChatGPT-3.5 demonstrated an evolving capacity to engage with complex medico-ethical dilemmas but also revealed biases and limitations. Policymakers must carefully integrate AI tools, incorporating broader economic and social insights while ensuring adaptability to diverse scenarios. The academic community and clinicians must remain vigilant and regulate the rapid implementation of AI in the increasingly uncertain and evolving healthcare landscape.