AGEISM IN ARTIFICIAL INTELLIGENCE: A REVIEW

Abstract

Artificial intelligence (AI) has emerged as a major driver of technol ogical development in the 21st century, yet little attention has been paid to algorithmic biases towards older adults. "Digital ageism" is a new form of ageism that is embedded into technology and AI systems. A im: This review aimed to explore how age-related bias is encoded in AI systems to better understand digital ageism. The scoping review follows a six-stage methodology framework develope d by Arksey and O'Malley. The search strategy has been established in six databases and we will investigate grey literature databases, targe ted websites, popular search engines. An iterative search strategy was used. Studies meet the inclusion criteria if they are in English, pee r-reviewed, available electronically in full-text, and included the co ncepts ‘bias’ and old age. At least two reviewers independently conduc ted title/abstract screening and full-text screening. Our database searches resulted in 7 595 manuscripts that underwent ti tle and abstract screening. Of these 49 papers, were included in the s tudy. The word "ageism" was explicitly mentioned only in about half of these papers. Approximately half the papers mentioned how age-related bias could be encoded into AI systems. The most commonly used AI appl icaiton was computer vision. Our preliminary findings contribute foundational knowledge about the age-related biases that were encoded or amplified in AI systems. This work advances how AI can be developed in a manner consistent with ethi cal values and human rights legislation, particularly as it relates to an older and aging population.