Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Exploring Alt Text Generation: A Comprehensive Study of Human and AI Approaches
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Web accessibility is a significant challenge for visually impaired individuals,specifically with the absence of meaningful alternative text (alt text) for images.According to accessibility guidelines, alt text is an essential part for screen readers todescribe visual content. However, its implementation is often overlooked orinconsistent. While AI based solutions have been proposed to help with the challengesand to automate the generation and implementation of alt text, its consistency andreliability remains uncertain. This study explores the potential of AI generated alt textby comparing it to human written alt text in terms of effectiveness and user satisfaction.The research was conducted through two surveys consisting of a mixed-methodsapproach. The first study gathered qualitative and quantitative data from visuallyimpaired individuals and their experience with alt text. The second survey was acomparative survey to compare visually impaired and sighted users’ preferencebetween human written and AI generated alt texts.The results indicate that while AI is limited in contextual awareness and is almostalways outperformed by human written alt text, it can assist in generating imagedescriptions. Visually impaired individuals prefer more contextual descriptions,whereas sighted users tend to choose more concise alternatives. The findings alsosuggest that AI can help developers, if integrated into their workflow for creating alttexts but should be reviewed and refined with human contextual input and awareness.Discussions conclude that AI can serve as a tool rather than a replacement for humanwritten alt texts, but it can be applicable in the context of a considerable number ofimages and is still better than no alt text at all or misuse of it.The research concludes that AI has the potential to contribute to the field of webaccessibility but is not yet fully reliable as a stand-alone solution. Further researchshould compare different AI models and their potential as well as to incorporatefeedback to create contextually aware AI models for alt text generation.

Place, publisher, year, edition, pages
2025. , p. 64
Keywords [en]
Web Accessibility, Alternative Text (Alt Text), Artificial Intelligence (AI), Visual Impairments, Survey
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:hj:diva-67642OAI: oai:DiVA.org:hj-67642DiVA, id: diva2:1954468
Subject / course
JTH, Computer Engineering
Supervisors
Examiners
Available from: 2025-04-29 Created: 2025-04-24 Last updated: 2025-10-13Bibliographically approved

Open Access in DiVA

No full text in DiVA

Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 213 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf