Creating Inclusive AI: A Strengths-Based Approach to Accessible Design
UX Strategies for Developing Empowering Systems for Diverse Users
The field of Artificial Intelligence (AI) has seen rapid advances in recent years, enabling new possibilities for intelligent systems that can adapt and personalise to individual users. However, these systems often fail to adequately consider the diverse needs and perspectives of people with varying cognitive abilities and disabilities. In this article, we will explore approaches researchers are taking to design more inclusive intelligent systems that empower users across the spectrum of human ability.
To start, we will compare deficit-based versus strengths-based perspectives on disability and how they inform the design of intelligent systems. We will then overview several key inclusive design frameworks that can guide the development of accessible and empowering systems. Next, we will discuss methodological considerations for conducting human-centred design research with diverse users. Finally, we will highlight insights and opportunities from recent studies for reframing intelligent systems to be more empowering and resonate with users' lived experiences.
Different Perspectives: Deficit vs. Strengths-Based Approaches
Many intelligent systems take a deficit-based perspective rooted in the medical model of disability. This views disability as an impairment or health condition within the individual that needs to be fixed or compensated for. From this perspective, systems attempt to identify and assess users' cognitive, physical, or sensory limitations, then provide features aimed at minimising or working around those deficits. For example, let’s think of an individual with ADHD. The medical model sees ADHD as a neurological deficit or disorder within the individual that needs to be fixed. It focuses on medicating symptoms or providing accommodations to help the person overcome the challenges of ADHD. Behaviours associated with ADHD, like distractibility and hyperactivity, as problems that need to be controlled or minimised. The aim of the intervention is “normalising” the affected person.
However, this approach risks further marginalising people by highlighting what they cannot do rather than leveraging their strengths. In contrast, the social model sees disability as the result of barriers in the environment and society, not inherent deficits in the person. Aligning with this view, strengths-based approaches focus on each person's unique abilities, interests, and lived experiences. Rather than fixating on deficits, they aim to understand the user holistically in context. Let’s consider ADHD again but from the social model perspective, which sees the condition as the result of environmental, social, and educational barriers rather than an inherent deficit in the individual. It focuses on identifying the strengths, talents, and interests the individual can leverage. The social model recognises traits associated with ADHD, like hyperfocus, divergent thinking, and high energy, as differences rather than deficits. It aims to accommodate the person's needs by modifying the environment, tasks, and expectations rather than trying to "normalise" the person.
But how can using each of those approaches affect design? Let’s answer this by looking an example. Imagine we’re working on a AI personal assistant. If it’s designed from a deficit perspective, it might only respond to voice commands. This limits access for users with speech impairments. A strengths-based approach would offer multiple interaction modes to leverage each user's capabilities, like voice, touchscreen, switch devices, or eye-tracking.
Key Frameworks for Inclusive Intelligent Systems
Several key frameworks guide the design of more accessible and empowering intelligent systems:
Universal Design aims to proactively consider diverse needs upfront rather than retrofitting accessibility later. It creates systems usable by all to the greatest extent possible without separate designs. For example, designing a voice assistant that also provides a text transcript caters to both deaf and hearing users from the start.
Ability-Based Design focuses on understanding users' spectrum of capabilities to tailor interfaces and interactions. This creates systems personalised to individuals or user groups. For instance, a memory aid app might offer multiple modes for entering and retrieving reminders based on a user's abilities.
Design for User Empowerment provides greater independence and dignity through intentional design decisions. This could involve enabling self-expression, adding transparency features, or increasing customisation to amplify users' agency. An example is allowing users to fully customise a smart home system's voice interface and responses.
Methodological Considerations
Quantitative methods like surveys and structured usability tests have limitations in capturing the nuanced, contextualised experiences of diverse users. While providing generalisable data, they do not facilitate deep exploration of each user's unique capabilities and interactional contexts.
Qualitative methods like interviews, observations, and co-design sessions allow for richer engagement with users and understanding of their perspectives. Working closely with users in context enables interpreting findings in alignment with their lived experiences. Key qualitative methods include:
Research Through Design: Building prototypes to investigate possibilities and reframe design directions
Co-Design: Enabling users to actively shape the design process
Long-Term Engagement: Developing holistic understanding of users over time
Reflexivity: Continuously examining one's own assumptions and biases
For example, researchers could conduct co-design sessions with users with aphasia where they collaboratively iterate on a speech-to-text app's interface and features using simple prototypes. This provides direct input from users' lived experiences.
Key Insights and Opportunities
Recent studies (Sitbon et al., 2023; Suijkerbuijk et al., 2023) point to several opportunities for reframing intelligent systems to be more empowering for diverse users:
Adopt strengths-based rather than deficit-based perspectives. Users' unique approaches can drive innovation.
Actively involve users through qualitative, co-design methods. Listen deeply and incorporate their feedback.
Sustain social connections during use, not just locate information. Conversations enabled have value beyond tasks.
Observe use in real-world contexts versus lab tests only.
Use iterative prototyping to sustain engaging dialogue with users over time. Users can help reshape systems.
Prioritise nurturing human connections and conversations versus notions of deficits.
In summary, taking an inclusive, flexible, and reflexive approach enables shaping intelligent systems that empower diverse users and resonate with their lived experiences. This facilitates more human-centred intelligent interactions.
Before I go… I have a personal update to share! I’d like to introduce you to my two new junior assistants, Alfred and Stanley Bonbon.
