AI Preference Optimization for Healthcare

We aim to run a head-to-head comparison of Kahneman-Tversky Optimization (KTO) vs Direct Preference Optimization (DPO) vs Reinforcement Learning from Human Feedback in learning preferences from community members around healthcare choices. The first application will be around vaccine hesitancy.

To optimize the deployment of AI in healthcare, it is crucial to involve community members directly. This integration not only mitigates bias but also enhances oversight, ensuring that AI applications align closely with patient needs. You would be working with physician scientists and researchers at UNC Gillings School of Public Health and Fred Hutchinson Cancer Research Center and this would likely result in a publication at an ICML workshop