Title : Implementing AI and Assistive Technologies for people with disabilities: Opportunities and challenges
Abstract:
Artificial intelligence is transforming the landscape for access and inclusion for disabled persons, bringing exciting opportunities and new challenges. In this session, we present the reality of the implementation of AI in the design and development of new assistive technologies to meet the needs of people with a disability. We recognise that Artificial intelligence (AI) is revolutionising traditional practices and enabling new capabilities. Whilst people with disabilities experience this change and disruption alongside the wider population, the changes may have more significant consequences for those with disabilities. An example is the use of AI as a core component for autonomous vehicles. Such innovation enhances the safety and use of vehicles for any citizen. However, for a person with a disability, these innovations bring the prospect of fully independent travel for the first time. Across multiple domains and sectors, we see how AI is making a difference. For instance:
•Assistive Technologies: AI-powered devices like voice assistants (e.g., Amazon Echo, Google Home) help individuals with mobility or speech impairments to control their environment, communicate, and access information more easily.
•Enhanced Communication: AI-driven applications, such as speech-to-text and text-to-speech software, enable people with hearing or speech disabilities to communicate more effectively. Eye-tracking and voice-recognition technologies also facilitate communication for those with severe physical disabilities.
•Personal Mobility: AI is revolutionising personal mobility through smart wheelchairs and navigation systems that help individuals with mobility impairments navigate their surroundings safely and independently.
•Inclusive Design: AI encourages the development of inclusive products and services that cater to a wide range of abilities, ensuring that technology is accessible to everyone.
However, the full implementation of inclusive AI for the whole population requires careful consideration of inclusive design built upon sound ethical principles and techniques. Experience has suggested that AI-based solutions can be severely biased against people with disabilities for several reasons. This would include a lack of diverse training data that amplifies historical bias. Ongoing design and development exclusion, algorithmic bias and a lack of accessibility standards: Addressing these issues requires a concerted effort to include people with disabilities in the AI development process, ensure diverse and representative training data, and implement robust fairness and accessibility standards. By doing so, we can create AI solutions that are more inclusive and equitable.
