AI and Digital Accessibility: A Bibliometric Analysis and Systematic Review

Digital accessibility refers to the practice of designing and developing digital systems, tools, and websites so that people with disabilities can use them. This encompasses individuals with visual, auditory, motor, or cognitive impairments, ensuring they can perceive, understand, navigate, and interact with digital content. The World Health Organization estimates that approximately 1.3 billion people, or 16% of the global population, experience significant disabilities, underscoring the societal need for accessible digital experiences. Beyond social responsibility, digital accessibility is crucial for legal compliance, fostering inclusion, and unlocking business opportunities. Artificial Intelligence (AI), on the other hand, is the simulation of human intelligence processes by machines, especially computer systems. These processes include learning, reasoning, and problem-solving. AI has seen remarkable advancements, particularly in machine learning, natural language processing, and computer vision, leading to applications like voice assistants, facial recognition, and predictive analytics. The integration of AI into our daily lives presents both opportunities and challenges for ensuring that these powerful technologies are accessible to everyone.

AI is actively being leveraged to create more accessible digital environments. Technologies such as Automated Speech Recognition (ASR) provide captions and subtitles for video content, benefiting individuals with hearing impairments. Image and facial recognition technologies can assist people with visual impairments by providing descriptions of visual elements. AI-generated summaries can break down lengthy texts into more digestible portions for screen reader users. Furthermore, AI can simulate user behavior to identify and resolve navigation issues, and automate regression testing for accessibility. AI-powered chatbots and virtual assistants offer alternative communication channels for individuals with speech and hearing impairments. Machine learning algorithms are also being developed to recognize patterns and identify accessibility barriers in digital content, such as automatically generating alternative text for images or correcting language that might be difficult for users with cognitive impairments to understand. The potential for AI to create more inclusive digital experiences is vast, provided these technologies are developed with careful consideration for diverse user needs.

The systematic review screened 3,706 articles, ultimately narrowing down to 43 relevant publications. These articles were analyzed and classified based on the defined framework. The research identified various AI applications contributing to digital accessibility, including those for artificial vision, navigation, educational tools, banking services, virtual assistants, and communication with smart devices. AI methodologies employed in these applications spanned computer vision, edge AI, Natural Language Processing (NLP), machine learning, and deep learning. The study also cataloged the design standards and frameworks used, as well as the challenges encountered in developing and implementing AI for accessibility.

The review underscored the critical importance of integrating accessibility standards and frameworks into the design and implementation of AI systems. While AI holds immense promise for enhancing digital accessibility, its effectiveness is significantly hampered when these systems fail to comply with established guidelines. This lack of adherence not only limits the usability for individuals with disabilities but also risks exacerbating existing inequalities. The research emphasizes the need for a fundamental shift in how AI solutions are developed, moving from a reactive approach to a proactive one where accessibility is a core design principle. This involves not only technical compliance but also a deeper understanding of the diverse needs and experiences of people with disabilities, ensuring that AI technologies are truly inclusive and beneficial for all.

This systematic review has provided a comprehensive analysis of the current state of AI applications in digital accessibility. The research highlights the significant potential of AI to enhance the lives of people with disabilities, but also identifies critical gaps and challenges. The predominant focus on visual impairments needs to be addressed, with increased research directed towards other disability types. Furthermore, a stronger emphasis on adhering to accessibility standards is crucial for developing truly inclusive AI systems. Future research should aim to bridge these gaps by exploring AI solutions for underrepresented disability groups, developing novel methodologies that prioritize accessibility by design, and fostering greater collaboration between AI developers and disability advocates. By adopting a more balanced and inclusive approach, we can harness the full potential of AI to create a digital world where everyone can participate fully and equitably.