Radiology Synthetic Confusion: How Generative Artificial Intelligence Amplifies Misunderstandings of Radiologists and...

Area of Science:

• Medical Imaging and Artificial Intelligence
• Healthcare Workforce Representation

Background:

• Generative AI tools are increasingly used for healthcare imagery.
• Distinguishing between radiologists and medical radiation technologists (MRTs) is often challenging for the public and providers.

Purpose of the Study:

• To evaluate the accuracy of generative AI in differentiating and depicting radiologists and MRTs.
• To identify biases in AI-generated healthcare role representations.

Main Methods:

• Assessed 1380 AI-generated images/videos from 8 text-to-image/video models.
• Five raters evaluated task-role accuracy, attire, equipment, lighting, isolation, and demographics.
• Statistical analysis compared model and role differences.

Main Results:

• Medical radiation technologists (MRTs) were accurately depicted in 82.0% of outputs; radiologists in only 56.2%.
• Inaccurate radiologist depictions often misrepresented MRT tasks (79.1%).
• Radiologist depictions showed demographic bias (more male, White) and attire/equipment inaccuracies compared to MRTs.

Conclusions:

• Generative AI frequently misrepresents radiologist roles, demographics, and tasks, perpetuating stereotypes.
• Current AI outputs can exacerbate public confusion between healthcare roles.
• Enhanced oversight and inclusion standards are necessary for equitable AI-generated healthcare content.