In a research experiment, ChatGPT, a large language model developed by OpenAI, has passed all three parts of the United States Medical Licensing Examination (USMLE), although just barely.
The USMLE is a standardized test that assesses the knowledge and skills of medical students and graduates, and is required for licensure to practice medicine in the United States. The test is divided into three parts: Part 1, Part 2 Clinical Knowledge, and Part 2 Clinical Skills.
Second-year medical students usually spend hundreds of hours preparing for Part 1, while Part 3 usually is taken by medical school graduates.
Ansible Health, a Silicon Valley startup focused on treating Chronic Obstructive Pulmonary Disease (COPD), had been researching various AI and machine learning tools to improve its care when they decided to test ChatGPT, the newest AI phenomenon.