Recapitulation of Ageism in Artificial Intelligence-Generated Images: Longitudinal Comparative Study

Area of Science:

• Artificial intelligence (AI) and its societal impact.
• Gerontology and the study of aging.
• Media representation and bias.

Background:

• Positive media portrayals of aging enhance older adults' health and longevity.
• Generative AI may amplify age-related societal biases, termed digital ageism.
• Research is needed on the evolution of digital ageism in AI images.

Purpose of the Study:

• To examine changes in visual characteristics of digital ageism in AI-generated older adult images over time.
• To understand the interaction between AI advancements, aging attitudes, and digital media's effect on older adults.

Main Methods:

• Longitudinal comparison of 164 AI-generated images from DALL-E 2 (2022 vs. 2023).
• Utilized identical geriatric lexicon text prompts for image generation.
• Evaluated demographic (gender, race, socioeconomic status) and emotional characteristics; analyzed using ANOVA.

Main Results:

• Predominantly White-racialized older adults depicted (5x higher than other races) in both years.
• Increased representation of Asian-racialized (20 to 31) and other racialized individuals (6 to 14).
• Consistent middle-class socioeconomic status representation; neutral emotional expressions in images despite neutral prompts.

Conclusions:

• AI image generators persist in creating digital ageism, despite positive societal aging narratives.
• Lack of improved emotional representation in AI images highlights persistent ageist visual characteristics.
• Findings underscore the need to address and mitigate digital ageism in AI-generated content.