Designing AI to Stop Disability Bias

The Team

The Proposal


Roughly 1 to 4 or 5 people in populations worldwide has a significant disability. AI, machine learning, automation, and new data technologies have considerable promise for addressing key barriers, challenges, and factors involved in the exclusion, discrimination, and lack of social and economic participation for people with disabilities.

Around the world, AI is being widely deployed, for instance, by companies such as Microsoft, Google, and others, as well as healthcare providers, social and community service providers, communication and media companies, information providers, to improve accessibility of technologies, goods, and services for people with disabilities. The potential of AI to advance the situation of people with disabilities, in line for instance, with the Sustainable Development Goals or the Convention on the Rights of Persons with Disabilities is a high priority, cross-cutting global. At the same time, evidence is emerging that some forms of AI and its design and deployment may be entrenching bias and discrimination against people with disabilities – or even being associated with new forms of exclusion and injustice.


As yet there is scant research available on AI and disability. The most systematic piece of research is the authoritative Nov 2019 report from AI Now Institute that lays out the key issues in disability, discrimination and bias (AI Now, 2019). The other leading work is the research of Prof Jutta Treviranus (Treviranus, 2018, 2019), who argues for inclusive data sets as key to responsible AI in disability.


Given the global state of play, this research addresses the problem statement identified by the research and debate to date, namely: how can we ensure responsible AI in relation to people with disabilities? (Clegg, 2020). Laying the ground for a scaled-up, comprehensive interdisciplinary study, this seed project is focussed on three key research questions:

  1. What concepts of disability are framing AI as it is developing in key areas of social life?
  2. What are the implications for current and emerging forms of AI for disability and social inclusion?
  3. What data sets are underpinning AI, machine learning, and automation in relation to disability? And how can we conceptualize, construct, and encourage more inclusive data sets?

The NTU team will be joined by international leaders as collaborators – Prof Calvo (Imperial) and Prof Treviranus (Inclusive Design Research Centre, OCAD, Toronto).


As a multi-ethnic, culturally diverse, technologically advanced society, where disability is receiving considerable recent attention Singapore offers an ideal ‘test-bed’ for investigating the ‘intersectional’ or complex ways that this aspect of AI is unfolding.Using a qualitative interdisciplinary approach drawing on disability studies, digital media research, science and technology studies, and HCI, we will focus on one strategic important area: AI in FinTech and its use by and implications for people with disabilities.

Our research plan is as follows:

  1. To establish the context, gather ‘best practice’ examples of AI in assistive technology developed globally and locally. Assistive technology tends to be a pioneering area in inclusive design for disability and accessibility, so this will help develop a ‘baseline’.
  2. Study the deployment of FinTech products and services, including via apps, in Singapore. Using a range of methods, including developer interviews, user interview and focus groups, and analysis of apps (using the ‘walk-through’ method, as well as critical disability studies methods), analyse:
  • how AI is being deployed in FinTech & whether AI design and implementation is being guided by objectives of inclusion, avoiding and overcoming bias and discrimination;
  • what the effects of AI in FinTech are for users with disabilities;
  • what data sets are being used to ‘train’ AI in emerging FinTech 

3. Formulate our findings in an accessible draft report for discussion at a small stakeholder workshop. 
4.  Finalize the report,and distribute. Prepare a paper on the research for publication in a top international journal. Prepare accessible article for publication in op-eds or widely read blogs.

Scale-Up Plan

This pilot project will provide a vital and invaluable stepping stone for scaling-up the research into a major interdisciplinary international project. Through the research project, we will consolidate and extend existing contacts with key Singaporean ministries and funding agencies, community organizations and service providers in disability, social services, health, and associated areas, with major international technology firms, and a wide range of technology developers and designers. We foresee an ambitious yet highly feasible landmark project that would undertake a comprehensive study of how to design AI to avoid disability bias across a range of strategic crucial sectors (workforce; education and training; health; banking and finance; and communication and media). Working with our collaborators, and other leaders in the field, to obtain such funding, we would target major international foundations as well as national research agencies.

Current Progress

We are currently undertaking research on FinTech and AI, to identity disability and accessibility implications – in order to refine and focus our approach.