Fostering websites accessibility: A case study on the use of the Large Language Models ChatGPT for automatic remediation

The use of automated accessibility testing tools remains a common practice for evaluating web accessibility. However, the results obtained from these tools may not always provide a comprehensive and complete view of a site's accessibility status. The main purpose of this study is to improve web accessibility by automatically remediating non-accessible ones using Large Language Models (LLM), particularly ChatGPT. The effectiveness of the used model in detecting and remediating accessibility issues to ensure compliance with the Web Content Accessibility Guidelines (WCAG 2.1) is also discussed. By using ChatGPT as a remediation tool, this study investigates the potential of LLM in improving web accessibility. In the case study, two websites that did not adhere to the WCAG 2.1 guidelines were selected as the primary experimental subjects for the study. These websites were assessed using the web accessibility evaluation tool, WAVE, to detect accessibility issues. The identified issues served then as the basis for remediation using ChatGPT. The effectiveness of the used advanced language model as a web accessibility remediation tool was evaluated by comparing its findings with those obtained from manual accessibility testing. The results of this comparison have significant implications for stakeholders involved in achieving WCAG compliance and contribute to the development of more accessible online platforms for individuals with disabilities.