Essay: Addressing technology bias against people with disabilities: Challenges and Solutions

November 1, 2024

James McKim,

Most of you know me as a champion of inclusion and equity from organizational and societal perspectives. But my background is in technology, and I have become increasingly interested in exposing where technology is failing not only the able-bodied but also the disabled. Let’s face it: We could all be disabled in one way or another someday.

In our increasingly digital world, technology plays a pivotal role in shaping how we live, work and interact. However, a concerning trend has emerged: Many technologies, both AI-powered and non-AI, exhibit biases against people with disabilities. This bias not only perpetuates existing inequalities but creates new barriers to inclusion and accessibility. As we continue to innovate, it’s crucial to address these biases and create technologies that truly serve all members of society.

The Problem of Bias in Technology
Recent studies reveal significant biases against people with disabilities in AI and machine- learning models. For instance, research from Penn State University shows that popular AI models exhibit explicit bias against disability-related terms in sentiment analysis and toxicity detection tasks.

However, the issue of bias extends beyond AI to many other forms of technology. User interfaces, hardware designs, websites and mobile applications often fail to consider the needs of users with various disabilities, creating significant accessibility barriers. These biases manifest in multiple ways:

Inaccessible digital interfaces that exclude users with visual, auditory or motor impairments.
Physical devices lacking inclusive design features.
Websites failing to meet accessibility standards.
Mobile apps without proper accessibility features.

Sources of Bias
Understanding the root causes of these biases is crucial for developing effective solutions. Some key factors contributing to technology bias include:

Lack of representation: Development teams often lack diverse perspectives, including those of people with disabilities. It’s like trying to bake a cake without flour.
Insufficient testing: Products may not be adequately tested with users who have disabilities.
Narrow design focus: Designers and developers may operate on limited definitions of disability or fail to consider the full spectrum of user needs. It’s like designing a one-size-fits-all sweater for a group of people with different body shapes.
Cost considerations: Accessibility features are sometimes viewed as expensive add-ons rather than essential components.

Consequences of Bias
The impact of these biases can be far-reaching and severely detrimental to people with disabilities:

Employment discrimination: AI-powered hiring tools and inaccessible job application systems may unfairly disadvantage candidates with disabilities.
Educational barriers: Students with disabilities may struggle with inaccessible learning management systems or digital textbooks.
Social exclusion: Biased content moderation systems and inaccessible social media platforms can isolate individuals with disabilities.
Health care disparities: Medical devices and health information systems that aren’t universally designed can lead to inadequate care.

Solutions to Mitigate Bias
Addressing these biases requires a multifaceted approach that encompasses both AI and non-AI technologies:

Inclusive design principles: Adopt universal design principles from the outset of product development to create technologies usable by all people to the greatest extent possible.
Diverse team composition: Include people with disabilities in technology development teams and decision-making processes to bring diverse perspectives to the table.
Comprehensive user testing: Implement rigorous testing protocols that include users with various disabilities to identify and address potential issues early in the development process.
Adherence to accessibility standards: Follow established accessibility guidelines, such as the Web Content Accessibility Guidelines (WCAG) for web content, across all digital products. Think of it as the “rules of the road” for the internet — everyone should know where to go.
Adaptive technologies: Develop and promote technologies that can be easily customized to meet diverse
user needs.
Education and training: Provide disability awareness and inclusive design training for technology professionals to raise awareness of potential biases and accessibility issues.
Legal and policy frameworks: Strengthen and enforce regulations requiring accessibility in technology products and services to ensure compliance and accountability.
Funding initiatives: Allocate resources specifically for developing accessible technologies and retrofitting existing ones to prioritize inclusivity.
Bias detection and mitigation in AI: Implement rigorous testing protocols to identify and address biases in AI models, particularly in areas like natural language processing and computer vision.
Transparency and accountability: Implement clear reporting mechanisms for algorithmic impact assessments and mitigation efforts to foster trust and enable continuous improvement.

Successful Examples of Bias Mitigation
While challenges remain, there are encouraging examples of non-AI technologies that have successfully mitigated bias:

Blind auditions in orchestras: The use of screens to conceal candidates’ identities during auditions has significantly increased the proportion of women hired by major symphony orchestras.
Anonymous resume-screening software: Tools that remove identifying information from resumes before human review have helped reduce unconscious bias in initial candidate screening for jobs.
Standardized testing accommodations: Technologies like screen readers and speech-to-text software have made standardized tests more accessible to people with disabilities. Because everyone deserves a fair shot.
Closed captioning and audio description: These technologies have improved media accessibility for people with hearing or visual impairments. Now everyone can enjoy the latest film.
Adjustable workstations: Height-adjustable desks and ergonomic equipment accommodate a wider range of body types and physical abilities in the workplace.

As technology continues to shape our world, it is crucial to address biases against people with disabilities across all forms of technology, both AI and non-AI. By implementing comprehensive solutions and maintaining a commitment to inclusivity, we can work towards creating a technological landscape that truly serves all members of society. The goal should be to harness the power of innovation to enhance accessibility and equal opportunity, rather than perpetuate existing barriers. Through collaborative efforts involving technologists, policymakers, disability advocates, and users, we can build a more inclusive digital future for everyone. After all, a world where everyone can participate is a world worth creating.

This article is featured in the winter 2024 issue of 603 Diversity.

603 Diversity’s mission is to educate readers of all backgrounds about the exciting accomplishments and cultural contributions of the state’s diverse communities, as well as the challenges faced and support needed by those communities to continue to grow and thrive in the Granite State.