Accessible Content Creation

Content That Everyone Can Use

🔄 Quick Recall: In the previous lesson, you learned to use AI for WCAG compliance auditing — running automated scans, prioritizing issues by impact, and integrating accessibility checks into CI/CD pipelines. Now you’ll apply AI to the content side: making the words, images, and media on your site accessible to everyone.

The biggest accessibility gap isn’t usually code — it’s content. Missing alt text, uncaptioned videos, unstructured documents, and complex language create barriers that no amount of clean HTML can overcome. AI makes accessible content creation practical at scale.

AI-Powered Alt Text

Writing alt text for every image on a website used to be one of the most time-consuming accessibility tasks. AI has transformed this — but understanding how to use AI effectively for alt text matters more than the tool itself.

The Alt Text Decision Framework

Help me write alt text for these images.
For each image I describe, tell me:

1. Which CATEGORY is this image?
   - Informative: conveys meaning → descriptive alt text
   - Decorative: visual only → alt="" (empty)
   - Functional: part of a link/button → describe the action
   - Complex: data/chart → short alt + long description

2. Draft the ALT TEXT following these rules:
   - 125 characters or fewer for standard alt text
   - Don't start with "image of" or "picture of"
   - Describe what the image COMMUNICATES, not just what it shows
   - Include context relevant to the page content
   - For product images: include name, key features, distinguishing details

3. Rate your CONFIDENCE (high/medium/low)
   - Low confidence means a human should prioritize reviewing this one

Images:
[describe each image and its context on the page]

Alt Text Examples

Image Description	Bad Alt Text	Good Alt Text	Why
Team photo on About page	“team”	“Five team members standing in the office lobby, smiling”	Conveys the scene, not just the subject
Line chart showing revenue growth	“chart”	“Revenue grew from $2M to $5M between 2023-2025”	Describes the data, not just the format
Search icon in a button	“magnifying glass”	“Search”	Describes function, not appearance
Decorative gradient background	“blue gradient”	alt="" (empty)	Decorative — should be invisible to screen readers

✅ Quick Check: Why is alt="" (empty alt attribute) different from a missing alt attribute? A missing alt attribute causes many screen readers to read the filename aloud (“DSC underscore zero eight four seven dot jpg”) — confusing and annoying. An empty alt attribute (alt="") tells the screen reader “this image is decorative, skip it.” For decorative images, empty alt is the correct, intentional choice.

AI-Generated Captions and Transcripts

Video Captioning Workflow

Help me create a captioning workflow for our video content.

We produce:
- [X] videos per month
- Average length: [X] minutes
- Content type: [tutorials/interviews/presentations/marketing]
- Budget for captioning: [limited/moderate/substantial]

Design a workflow:

STEP 1 — AI AUTO-CAPTION:
- Tool recommendation based on our needs and budget
- Expected accuracy rate for our content type
- Handling of technical terms and proper nouns

STEP 2 — HUMAN REVIEW:
- Which errors to prioritize (names, numbers, technical terms)
- How to handle multiple speakers
- Adding non-speech audio descriptions [applause], [music]
- Timing synchronization check

STEP 3 — TRANSCRIPT CREATION:
- Full-text transcript from corrected captions
- Add speaker labels and timestamps
- Include visual descriptions for audio-only consumers
- Format for web publication (searchable, scannable)

STEP 4 — QUALITY CHECK:
- Accuracy target: 99%+ for published captions
- Timing: captions within 1-2 frames of speech
- Readability: max 2 lines, 42 characters per line

Caption Quality Standards

Element	Standard	Common AI Error
Accuracy	99%+ for published content	Proper nouns, technical terms, numbers
Timing	Within 1-2 frames of speech	Lagging in fast-paced dialogue
Line length	Max 42 characters per line	Run-on captions that are hard to read
Speaker ID	Labeled for multi-speaker content	Fails to identify speaker changes
Non-speech audio	[music], [applause], [phone ringing]	Typically missing in AI-generated captions

Accessible Document Creation

Structured Documents

Help me make this document accessible.

Content: [paste document content or describe it]
Format: [Word/PDF/Google Doc/web page]
Audience: [internal/public/regulatory]

Apply these accessibility requirements:

STRUCTURE:
- Proper heading hierarchy (H1 → H2 → H3, no skipped levels)
- Meaningful headings (descriptive, not "Section 1")
- Logical reading order

IMAGES:
- Alt text for all informative images
- Empty alt for decorative images
- Long descriptions for complex charts/diagrams

TABLES:
- Header rows and columns marked
- Simple structure (avoid merged cells if possible)
- Caption describing what the table shows

TEXT:
- Plain language (target 8th grade reading level)
- Sufficient color contrast
- Links with descriptive text (not "click here")

FORMATTING:
- Use lists for grouped items (not manual bullets)
- Use styles for emphasis (not just color)
- Consistent layout and navigation

The Plain Language Principle

Rewrite this content in plain language for accessibility.

Original: [paste content]

Requirements:
- Target reading level: 8th grade (Flesch-Kincaid)
- Short sentences (under 25 words average)
- Common words (avoid jargon without explanation)
- Active voice preferred
- One idea per paragraph
- Define technical terms on first use

Keep: all the factual content and nuance
Remove: unnecessary complexity, passive constructions,
  and abstract language that doesn't add meaning

Show me the before/after with the reading level
score for each version.

✅ Quick Check: Why does plain language benefit ALL users, not just those with cognitive disabilities? Because cognitive load affects everyone — complex language slows comprehension for all readers, especially when they’re tired, distracted, reading in a second language, or scanning on a mobile device. Research shows plain language improves comprehension by 50-80% across all literacy levels. It’s not “dumbing down” — it’s removing unnecessary barriers to understanding.

Key Takeaways

Categorize images before writing alt text: informative (describe meaning), decorative (alt=""), functional (describe action), and complex (short alt + long description)
AI generates alt text drafts in minutes, but human review catches context errors — AI describes what it sees, not what it means in context
Video captioning follows a workflow: AI auto-caption → human review (names, numbers, speaker ID) → non-speech audio descriptions → quality check targeting 99%+ accuracy
Document accessibility starts in the source (proper heading styles, table markup, alt text) — not in the exported PDF
Plain language (8th grade reading level) improves comprehension for all users by 50-80%, not just those with cognitive disabilities

Up Next: You’ll explore cognitive accessibility and neurodiverse design — using AI to simplify content, support executive function, and create adaptive interfaces for users with ADHD, dyslexia, autism, and other cognitive differences.