York-led international workshop examines data bias through disability lens

Three people looking at a laptop screen

A two-day workshop co-organized by York University researchers brought local, national and international disability organizations together to discuss access to disability data to address human rights, disability justice and Sustainable Development Goals (SDGs).

The event, titled “Disability Sector Strategies for Addressing Data Needs, AI-Bias and Data Ableism,” ran May on 21 and 22 as part of the research team’s Social Sciences and Humanities Research Council of Canada (SSHRC) Insight Development Grant.

The SSHRC project involved the design and implementation of a virtual community platform using Wikibase, Semantic Web, machine learning and web programming tools, to enable disability communities to upload and search for disability documents. The platform data model is based on the United Nations Convention on the Rights of Persons with Disabilities (CRPD). The virtual community facilitates the uploading and sharing of validated information and supports disability rights advocacy by enabling the dissemination of knowledge.

Christo El Morr
Rachel Da Silveira Gorman
Rachel Da Silveira Gorman

Researchers on the project are led by York University Faculty of Health Professor Christo El Morr and Associate Professor in the Critical Disabilities Studies program (CDS) Rachel da Silveira Gorman, who organized the two-day workshop with their French collaborators Pierre Maret and Fabrice Muhlenbach from the University of Saint Etienne, as well as York colleagues Thumeka Mgwigwi and Serban Dina-Panaitescu. Alexis Buettgen (PhD graduate from CDS) was a collaborator on the project and Alexandra Creighton from the Health Equity program (PhD student in CDS) and Bushra Kundi (Health Studies program) were the research assistants.

Gorman introduced the theme of the first day – Disability data needs: Piloting disability justice data strategies – and outlined why data ableism and artificial intelligence (AI)-bias matter to local and international disability communities. El Morr presented the Disability Wiki project aims and progress, while Maret detailed the process of building a disability justice-focused AI search engine for the Disability wikibase pilot. Angela Stanley, a PhD candidate in Gender, Feminist and Women’s Studies, presented on digital literacy and accessibility, and tailoring accessibility to meet the needs of diverse disability communities.

Over the two days, participating disability organizations discussed their data needs and challenges in accessing data. Discussants included Yasmine Gray (Disability Justice Network of Ontario, and CDS MA student), Michaela Knot (Canadian National Institute for the Blind), Sam Walsh (Disabled Women Network Canada), Viviane Lee (Information and Communications Technology Council), Daniella Levy Pinto (National Network for Equitable Library Service), Angela Stanley (Ontario Digital Literacy and Access Network), Megan Linton, Chris Rowley and Catherine Rodgers (People First Canada), and Paula Hearn (International Disability Alliance, and CDS MA graduate).

“The use of artificial intelligence can lead to positive impact, but it has also the potential to propagate or embolden already existing biases,” said El Morr. “Our multidisciplinary research with disability advocacy groups, international collaborators, and technology partner (The QA Company) can help to address some of the existing biases in disability data.”

Gorman explains that “most data about disability are produced by medical, educational and social services sectors, and are framed by medicalized and deficit models. There is a lack of high quality and comparative data to support disability justice initiatives and disability rights claims.”

The second day addressed the theme Resisting AI and data bias: Strategies for community-led data justice. Muhlebach presented the data mining lifecycle, the importance of understanding the problem and the data, and examples of data bias. El Morr presented the types and categories of AI biases including social, racial, gender, and intersectional bias and their impact through examples. Annalise Clarkson, a PhD student in Critical Disability Studies, presented on the topic of data democratization and the importance of having access to high-quality data.

Invited researchers Jacqui Getfield (scholar on Black mothering and disability justice), Professor Evelyn F. Kissi (College of Education, University of Ghana), De-Lawrence Lamptey (Provostial Postdoctoral Fellow, York University), and Yvonne Simpson (curriculum development consultant, Toronto Metropolitan University) spoke on data bias, anti-Black racism and disability.

Participants at Disability Sector Strategies for Addressing Data Needs, AI-Bias and Data Ableism

Over the two days, researchers and organizations learned that health informatics and AI that have been developed through an equity-focused process provide important tools that can help provide access to disability data.

During the next phase of the project, the partners will work with local, national and international partners to pilot the Disability Wikibase and search algorithm and develop disability justice-focused and anti-racist toolkits on disability data and data sovereignty.

“In disability movements, many disabled people are under-represented based on gender and gender identity, race, citizenship status, and institutionalization,” said Gorman. “Many of our disability sector partners are eager to address the lack of representation of Black and Indigenous people and people of colour within their organizations.”