24/04/2026
Last year, I joined a group of physicians working on a mission as ambitious as it is necessary: improving how AI performs in real-world clinical settings.
I’m proud to have contributed, alongside a world-class international team, most advanced medical model: ChatGPT for Clinicians.
We’re also supporting this launch with the paper "HealthBench Professional: Evaluating Large Language Models on Real Clinician Chats", which introduces an open and highly challenging benchmark based on real clinician conversations.
One key finding stands out: in this deliberately demanding setting, ChatGPT for Clinicians not only outperforms other evaluated models (see chart), but also surpasses physicians on the same tasks.
What makes this work particularly meaningful is the rigor behind it—grounded in real clinical complexity, where decisions are rarely straightforward and always carry responsibility.
All of this would not be possible without this incredible team, along with Rebecca Soskin Hicks, MD leadership and Paola Rodríguez - MD, Eng, MSc. excellent team management!
This is just the beginning, but it’s a strong signal of what’s possible when clinicians are truly at the center of innovation.
➡️ Link to the official OpenAI announcement and full research paper in bio
