Balancing Safety and Helpfulness in Healthcare AI Assistants through Iterative Preference Alignment

Open in new window