A study from Penn State found ChatGPT’s 4o model produced better results on 50 multiple-choice questions as researchers’ ...