Mahmud Omar

Hi all, my name is Mahmud, I'm a physician, and a researcher focused on clinical genAI, more specifically of testing and ensuring the models work to make equity, safety and assurance better and not worse :)

Most of my working hours (other than the clinic of course) is figuring out what language models do, and don't do, when you put them next to medicine. For example, most of what I publish lives around a handful of questions. What clinical recommendations LLMs hallucinate, and why. How their advice shifts when the patient's identity is named, or the same clinical case is worded differently. What happens when AI agents move from answering questions to working with patient records. My aim, or really the dream is, to get continuous equity and safety driven evaluation, hard-carried guardrails that test equity, debiasing, diversity in data presentation and safe, not unwarrantedly varied LLM outputs in medicine. In simple terms: the AI gives the accurate, yet personal answer, in medicine for you, no matter who you are, what language you talk, how you look, and which are your preferences (other than not finding Better Call Saul as the best series ever, this is not acceptable).

In my research work, I'm currently working as the Head of Research at the BRIDGE GenAI Lab in BIDMC (Beth Israel Deaconess Medical Center, Harvard Medical School), and work as a research scientist at the Windreich Department of Artificial Intelligence and Human Health, at Mount Sinai.

SO, if this talks to you, and you think you and I can collab, really collab on actionable unique and novel ideas, I think we might connect (maybe we'll talk also, among others, about epic fantasies? books? lego? serieses? and beautiful nature places in the world?)