AI and the Future of Accessible Work
Anne Marie Piper, Darren Gergle, Cynthia L. Bennett, Maitreya Shah, Chancey Fleet, Ali Abdolrahmani, Ariana Aboulafia, Stacy M. Branham, Sharon Koppman, Arielle Silverman, Erik B. SudderthWhile AI offers exciting potential to improve workplace accessibility through real-time captioning, image alt-text generation, and writing support, it also poses significant risks and can create new barriers for disabled workers. Grounded in discussions among experts in accessibility, work and organizations, AI and machine learning, and disability policy, this article identifies six key consequences of AI for accessible work. These consequences are not merely technical, but are also structural, organizational, and social, and emotional, creating wide-ranging impacts on disabled workers. The use cases, norms, and concerns of AI for the future of work are rapidly evolving, creating new considerations for disabled workers who may simultaneously depend on and deplore AI tools. While we focus in particular on issues experienced by blind and low-vision workers in the U.S., we surface concerns that are likely to impact other disabled communities within and outside of the U.S. We conclude with four recommendations for moving toward responsible AI for accessible work, which articulate opportunities for future research, workplace accommodations processes, auditing of AI technologies, and AI governance and policy.