Editable Alternatives Research Proposal

Hi there,

Regarding Supplemental requirement: Editable alternatives. This reflects
recognition that auto‑generated text, though useful, often needs human
adjustment. WCAG 3.0 is designed to incorporate ATAG 2.0 principles
(Authoring Tool Accessibility), particularly in giving authors full control
over accessibility content. Making auto-generated text descriptions
editable by the content creator is both sensible and supported by WCAG 3.0
framework, allowing tools to provide a strong starting point while ensuring
human oversight refines meaning and context. The proposal aligns with
emerging accessibility standards and addresses practical usability concerns..

To validate it, key research should focus on end-user comprehension,
accuracy of content post-editing, collaboration workflows with content
authors, and ethical integrity in outputs. Together, this can help
determine whether this approach should move from “supporting requirement”
to a formal requirement in future guideline versions. Expanded Proposal:
Editable Auto-Generated Text Descriptions in WCAG 3.01. Executive Summary

This proposal recommends the adoption and formalisation of editable
auto-generated text descriptions within WCAG 3.0 and related authoring tool
accessibility guidelines. The purpose is to harness the efficiency of
AI-generated descriptions while enabling human content creators to refine
and contextualise them, resulting in more accurate, relevant, and
user-focused content. The proposal aligns with WCAG 3.0’s focus on
usability and inclusivity and addresses key issues of current automated
approaches, such as inaccuracies, lack of context, and ethical concerns.
2. Background & Context

WCAG 2.x requires meaningful text alternatives for non-text content but
does not directly address AI-generated descriptions or their editability.
Many modern content management systems and platforms use AI-based image
recognition to create alt text but do not offer an option for the author to
edit these descriptions before publishing. This can result in misleading or
incomplete descriptions that meet minimum technical requirements but fail
to deliver true accessibility. WCAG 3.0, in its draft form, includes
editable auto-generated text as a supporting requirement, acknowledging its
potential to improve accessibility outcomes.
3. Proposed Solution

The proposed solution is to integrate editable AI-generated descriptions
directly into content creation workflows. AI would generate an initial
description based on the image or non-text content, and the content creator
would be able to review and refine the description before it is published.
The editing interface would provide:
• Inline editing options directly within the content management environment.
• Version comparison to view AI-generated and edited descriptions
side-by-side.
• Contextual prompts and WCAG-aligned suggestions to guide improvements.
4. Visual Workflow Overview

1. Image uploaded to content platform.
2. AI auto-generates initial text description.
3. Content creator reviews description in inline editor.
4. Creator edits text based on context, accuracy, and WCAG guidelines.
5. Final description saved and published alongside the image.
5. Benefits

• Accuracy & Relevance: Human review ensures descriptions are contextually
correct and meaningful.
• Efficiency: Reduces authoring time while maintaining high quality..
• Enhanced User Experience: Improves clarity and comprehension for
assistive technology users.
• Compliance: Stronger alignment with WCAG 3.0 and ATAG 2.0.
• Ethical Safeguards: Reduces risk of bias or misrepresentation in
automated descriptions.
6. Risks & Mitigations

Risk: Authors may bypass editing.
Mitigation: Implement mandatory review prompts before publishing.

Risk: Varied quality of edits.
Mitigation: Provide training and quick WCAG editing reference materials.

Risk: Over-dependence on AI.
Mitigation: Ensure that AI text is clearly marked as 'machine-generated'
until edited.
7. Research Plan

A. Comparative Usability Testing: Test user experience with AI-only,
editable AI, and human-written descriptions.
B. Workflow Efficiency Studies: Track author time savings and perceived
value.
C. Accuracy Analysis: Quantify improvements after editing using error
categorisation.
D. Contextual Effectiveness Testing: Validate AI plus human-edited text in
domain-specific scenarios.
8. Pilot Implementation Plan

1. Select pilot CMS or authoring tools.
2. Enable editable auto-generated descriptions.
3. Train authors in WCAG-compliant editing.
4. Monitor edits and collect data over 3–6 months.
5. Evaluate based on accessibility compliance, accuracy, and user
satisfaction metrics.
9. Measurement & Success Indicators

• Reduction in accessibility errors.
• Improved end-user comprehension and satisfaction.
• Increased adoption of accessibility features.
• Positive feedback from both authors and end users.
10. Conclusion

Editable auto-generated descriptions represent a practical, impactful way
to enhance digital accessibility. This approach preserves the speed of AI
generation while ensuring human oversight delivers contextual accuracy and
better usability. Adoption within WCAG 3.0 will set a higher standard for
accessibility authoring tools and help bridge the gap between compliance
and true inclusion.


Regards

Karl Bernoth