Beyond the stereotypes: Artificial Intelligence image generation and diversity in anesthesiology

• DOI: 10.3389/frai.2024.1462819
• Publication Date: 2024-10-09T13:57:59Z
• Authors: Mia Gisselbaek; Laurens Minsart; Ekin Köselerli; Mélanie Suppan; Basak Ceyda Meco; Laurence Seidel; Adelin Albert; Odmara L. Barreto...; Sarah Saxena; Joana Berger-Esti...
• Keywords: IMG

Introduction Artificial Intelligence (AI) is increasingly being integrated into anesthesiology to enhance patient safety, improve efficiency, and streamline various aspects of practice. Objective This study aims evaluate whether AI-generated images accurately depict the demographic racial ethnic diversity observed in Anesthesia workforce identify inherent social biases these images. Methods cross-sectional analysis was conducted from January February 2024. Demographic data were collected American Society Anesthesiologists (ASA) European Anesthesiology Intensive Care (ESAIC). Two AI text-to-image models, ChatGPT DALL-E 2 Midjourney, generated anesthesiologists across subspecialties. Three independent reviewers assessed categorized each image based on sex, race/ethnicity, age, emotional traits. Results A total 1,200 analyzed. We found significant discrepancies between actual data. The models predominantly portrayed as White, with DALL-E2 at 64.2% Midjourney 83.0%. Moreover, male gender highly associated White ethnicity by (79.1%) non-White (87%). Age distribution also varied significantly, younger underrepresented. revealed predominant traits such “masculine, ““attractive, “and “trustworthy” Conclusion exhibited notable gender, age representation, failing reflect within anesthesiologist workforce. These highlight need for more diverse training datasets strategies mitigate bias ensure accurate inclusive representations medical field.