Edit

Hippocratic AI

https://www.hippocraticai.com/

Hippocratic AI is building a safety focused large language model (LLM) for the healthcare industry. We believe that generative AI has the potential to massively increase healthcare access the world over but has to be built and tested responsibly. Like the Hippocratic oath that doctors take we are building a model that aims to “Do no Harm.” Hippocratic AI’s model is tuned for the healthcare industry in three ways: 1. Certifications: It was trained to pass not just the USMLE but rather to pass 114 different detailed certifications. Furthermore, beat open AI’s GPT-4 on 105 of the certifications (tied on 6 and only lost on 3) 2. Bedside Manner: Specifically trained to have a great bedside manner. In healthcare delivery, how you say something to a patient is as important as what you say. The company develop a novel benchmark and beat all other commercial LLMs on this measure of bedside manner. 3. RHLF’d by Healthcare Professionals: RLHF is the process of tuning a model and is a technique credited with must of the success of ChatGPT. Hippocratic is using the exact healthcare professions that conduct each role today to QA and train the model. The company raised a $50MM seed round from is A16z and General Catalyst and is founded by hospital administrators, doctors, medicare experts, and LLM researchers.