Artificial Intelligence Portrayals in Orthopaedic Surgery: An Analysis of Gender and Racial Diversity Using...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Societal Perceptions

Background:

• Increasing accessibility of AI text-to-image generators.
• Novel avenue for exploring societal perceptions.
• Assessed AI-generated images for orthopaedic surgeon diversity.

Purpose of the Study:

• Examine gender and racial diversity in AI-generated orthopaedic surgeon images.
• Compare AI representation with the American Academy of Orthopaedic Surgeons (AAOS) census.
• Identify potential biases in AI portrayal of medical professionals.

Main Methods:

• Utilized five AI text-to-image generators (DALL·E 2, Runway, Midjourney, ImagineAI, JasperArt).
• Generated 240 images per generator using specific search queries.
• Two reviewers categorized demographics; compared with 2018 AAOS census data.

Main Results:

• AI images predominantly depicted White (82.8%) male (94.5%) surgeons, often older (64.4% ≥50).
• DALL·E 2 showed increased diversity in women and non-White surgeons compared to AAOS census.
• Midjourney, Runway, and ImagineAI showed significantly decreased diversity.

Conclusions:

• AI image generation shows distortions in orthopaedic surgeon diversity.
• Potential to influence public perception and reinforce healthcare disparities.
• DALL·E 2 and JasperArt show promise, but improvements needed in other AI tools for accurate, inclusive representation.