AI-enabled AT Framework: a principles-based approach to emerging assistive technology.

Abstract

Purpose: Assistive Technology (AT) is an umbrella term that describes the combination of devices and services used by individuals with a disability to perform tasks that might otherwise be difficult or impossible to complete due to their disability. Increasingly, Artificial Intelligence (AI) is being used in the development of innovative AT. Given the diverse applications of AI and the unique needs of people with disability, a practical approach that facilitates informed decision-making for all stakeholders while supporting choice and control for people with disability, in the AI-enabled AT space, is essential. This paper presents the ''AI-enabled AT Framework'', a tool that aims to facilitate effective decision-making, development, and assessment of AI-enabled AT.

Materials and Methods: The framework was co-designed through a participatory research approach, engaging key stakeholders, including people with disabilities, carers and support people, AI and AT industry representatives, government bodies, and researchers. A multi-stage process was employed, including literature review, interviews, focus groups, and industry workshops.

Results: The AI-enabled AT Framework provides a structured, person-centered approach for assessing AI-enabled AT, incorporating six core domains: user experience, privacy and security, quality, safety, relative value, and human rights. It supports decision-making for stakeholders by providing clear evaluation criteria to assess AI-enabled AT. The framework highlights the importance of ongoing stakeholder engagement and outlines a roadmap for implementation, refinement, and adoption.

Conclusion: The AI-enabled AT Framework offers a practical tool to enhance decision-making in the development, evaluation, and deployment of AI-enabled AT. By emphasizing co-design and stakeholder engagement, it promotes ethical, effective, and user-centered AI applications. Future research should focus on framework validation, implementation strategies, and addressing emerging challenges in AI-enabled AT adoption.