Researchers evaluated how well AI models GPT-4 and ChatGPT perform on the US Medical Licensing Examination questions focused on soft skills like judgment and empathy. GPT-4 outperformed both ChatGPT and human test-takers, answering 90% of questions correctly, while ChatGPT had a 62.5% accuracy rate.
from News Medical Medical Research News Feed https://ift.tt/iw4DPF7
0 Comments