Abstract

As accessibility legislation becomes increasingly stringent, publishers are turning to AI to help meet their obligations. However, the integration of AI into accessible publishing presents significant challenges. While AI tools show promise, their application in accessibility remediation is far from straightforward and requires careful consideration. Tasks such as providing image descriptions, creating accessibility metadata, and structuring content for assistive technologies require complex, subjective editorial decisions that AI currently struggles to handle effectively. AI’s limitations in contextual analysis and its inability to fully replicate human judgment mean that its use for these tasks is not yet reliable without substantial human oversight. Additionally, issues like copyright risk and the need for tailored content checks further complicate AI's role in accessibility. This article explores the current state of AI in accessible publishing, emphasizing the need for cautious, informed implementation and further research to assess its true potential in this critical area.

Keywords

Accessibility, Content Creators, Large Language Models

How to Cite

Rights

© 2024 Simon Mellins. This article is distributed under the terms of the Creative Commons Attribution Licence (CC-BY) 4.0 https://creativecommons.org/licences/by/4.0/, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.