Age Representations in Generative AI Art from MidJourney

Abstract

This dataset of images was analyzed in a study examining how Midjourney, an Artificial Intelligence platform which uses text to generate images, visually portrays older and younger adults. A mixed-methods approach was applied to 456 images generated in Midjourney. The 27 sub-dimensions of the old age exclusion framework by Walsh, Scharf, and Keating (2017) were used as pilot searches in Midjourney to generate four images each. The research team reviewed the images for coherence and revised the list, totaling 19 keywords for the final searches. Each keyword was paired with “old person” for three searches (totaling 12 images) and “young person” for three searches (totaling 12 images), for a grand total of 456 images. The following is the list of keywords: citizenship; general civic activities; volunteering and community responsibility; voting and political participation; healthcare services; transport and mobility; information and ICT; housing; social networks and support; relationship; social opportunities; social participation; sidewalk; physical infrastructure; neighborhood and community; socio-economic status; material resources; common consumer goods.