THE AI CHAPTER OF DISABILITY EMPLOYMENT

Written by Mary Bellard, Principal Innovation Architect Lead at Microsoft

When looking ahead to the future of work and the workplace, AI’s role will continue to evolve. And when we acknowledge that the employment rate of people with disabilities is less than half that of people without a disability, despite the availability of disability talent and the societal benefits we all experience because of accessibility, we have to ask ourselves – what role can AI have in bridging the disability divide to increase disability employment?

As the shift to remote work during the COVID-19 pandemic has disproportionately impacted people with disabilities, the importance of accessible, intelligent productivity tools has never been more important. From generating captions with automatic speech recognition (ASR) in Teams, to recognizing objects with computer vision in Seeing AI, to leveraging the Immersive Reader service on Azure, AI continues to show promise for empowering people to do more. To make AI inclusive, however, we need to think not just about what AI can do but how it does it, too.

Microsoft launched the AI for Accessibility program in 2018 to help innovators harness the potential of AI to empower the more than one billion people around the world with a disability, and employment remains a focus area of the program. Through our grantee engagements, we have come to understand that we are in a data desert for disability-specific data from which to build more powerful and meaningful machine learning systems. This has guided Microsoft to invest in more projects that better represent the disability experience such as datasets of images submitted directly by people who are blind or have low vision to build more accurate description models. The gap for inclusive datasets that represent disability is shared across many domains as described in the AI Now Institute at NYU’s Report on Disability, Bias, and AI.

We need the expertise and creativity of people with disabilities to shape what inclusive AI can be, and to be part of the conversations and projects that define how technology impacts key aspects of society such as the classroom and the workplace.

October marks National Disability Employment Awareness Month (NDEAM) in the U.S., to honor and celebrate the many accomplishments people with disabilities have made as valuable members of the workforce. It is a critical reminder that who builds inclusive AI will also determine the extent of inclusion. If you are building AI tools – consider the representation of disability in the datasets you are working with. If you hire for AI projects – recruit disability talent. And if you are a person with a disability already working in AI – you are actively shaping this next chapter of technology and the (r)evolution of inclusive AI will be stronger because of you.