Scholarlink

<p>This article investigates the extent to which AI-generated descriptions can enhance the accessibility of business graphics for individuals who are blind or visually impaired. Specifically, it examines which AI models currently integrated into screen readers demonstrate the highest levels of performance, as well as the limitations and advantages associated with the use of such descriptions.</p><p>The research questions are addressed through a&#xa0;systematic review of selected scholarly literature and an empirical performance analysis of relevant AI models. Furthermore, survey data collected from blind and visually impaired participants are incorporated and evaluated as part of an assessment of the proposed approach.</p><p>The findings suggest that, among the models examined, Gemini&#xa0;2.0 Flash and Claude&#xa0;3.5 Sonnet currently achieve the most effective results in the description of business graphics. Nevertheless, notable shortcomings remain, particularly with respect to the accurate and objective representation of data points. These limitations are accompanied by a&#xa0;persistent lack of user trust in the reliability of AI-generated descriptions. At the same time, the study highlights the potential benefits of employing AI models, most notably the reduction of accessibility barriers and the promotion of more inclusive practices.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Künstliche Intelligenz zur Verbesserung der Zugänglichkeit von Business-Graphiken für Menschen mit Sehbeeinträchtigungen

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月2日

查看全文
  • Selina Hölzl,
  • Christian Langenbach

摘要

This article investigates the extent to which AI-generated descriptions can enhance the accessibility of business graphics for individuals who are blind or visually impaired. Specifically, it examines which AI models currently integrated into screen readers demonstrate the highest levels of performance, as well as the limitations and advantages associated with the use of such descriptions.

The research questions are addressed through a systematic review of selected scholarly literature and an empirical performance analysis of relevant AI models. Furthermore, survey data collected from blind and visually impaired participants are incorporated and evaluated as part of an assessment of the proposed approach.

The findings suggest that, among the models examined, Gemini 2.0 Flash and Claude 3.5 Sonnet currently achieve the most effective results in the description of business graphics. Nevertheless, notable shortcomings remain, particularly with respect to the accurate and objective representation of data points. These limitations are accompanied by a persistent lack of user trust in the reliability of AI-generated descriptions. At the same time, the study highlights the potential benefits of employing AI models, most notably the reduction of accessibility barriers and the promotion of more inclusive practices.