Heuristic Evaluation of AI-Powered Web Accessibility Assistants

Web-based services occupy an increasingly large place in the lives of people around the world. Users with disabilities may find it difficult to use many websites and platforms due to technical barriers. To improve access to websites, international organisations like International Organization for Standardization (ISO), European Telecommunications Standards Institute (ETSI), or World Wide Web Consortium (W3C) develop web accessibility standards and guidelines, which provide an introduction to accessibility issues. By following some basic recommendations, any website and platform can become a good environment for exchanging information to digitally include people with disabilities. Due to the rapid development of artificial intelligence (AI) recent 5 years and its application in various sectors (e.g., game industry, e-commerce, facial recognition systems, agriculture, financial and economic sectors, cyber security, education, etc.), this also implies its application to help people with special needs to ensure digital accessibility. In this regard, the aim of this paper is related to proposing a method for heuristic evaluation of AI powered web accessibility assistants. The method was tested using AI-powered accessibility assistants. The paper's objectives that meet its purpose and research questions are: (1) review of artificial intelligence application in web accessibility practices; (2) evaluation of selected tools based on the proposed method to prove its applicability. The method has been tested by heuristic evaluation of AI-powered accessibility assistants for websites. Web accessibility, measured in number of errors and alerts by applying WAVE tool, was compared before and after accessibility assistants were enabled.