ChatGPT's medical diagnoses are accurate less than half of the time, a new study reveals.

Scientists asked the artificial intelligence (AI) chatbot to assess 150 case studies from the medical website Medscape and found that GPT 3.5 (which powered ChatGPT when it launched in 2022) only gave a correct diagnosis 49% of the time.

Previous research showed that the chatbot could scrape a pass in the United States Medical Licensing Exam (USMLE) — a finding hailed by its authors as "a notable milestone in AI maturation."

Gebhard Sammer and Annette Kroiß (University Gießen) present in ´Psychological Test and Assessment Modeling´ a new picture-based method that serves to capture the anticipation of social conflicts based on perception.

"Examinations commonly use either free-response (FR) or multiple-choice (MC) response formats, each with its advantages and disantvantages. Despite frequent reports of high construct equivalence between them, certain group differences based on differing person characteristics still need to be explained." Prof. Dr. Tuulia M. Ortner et al. "aimed to investigate how test takers´ characteristics and behavior - particularly test anxiety, risk propensity, conscientiousness, lecture attendance, and study time - impact test scores in exams with FR and MC format." The authors report in ´Psychological Test and Assessment Modeling´ (1/24):

Managers in international companies need to be transculturally competent to realise synergies available from cultural diversity and to communicate and integrate effectively across differences, Prof. Claude-Helene Mayer and Dr. Christian Martin Boness emphasize in their manual „Creating mental health across cultures – Coaching and training for managers“: