Prompt Engineering for Accessibility: Making AI Do What You Need

A prompt is simply the instruction you give an AI tool. Prompt engineering means writing that instruction in a way that gets you exactly what you need. In accessibility work, the difference between a vague prompt and a well-crafted one can mean the difference between useful output and something you have to redo.

Think of it like asking for directions. "Tell me how to get there" is vague. "Give me turn-by-turn directions, avoiding highways, that work for someone who uses a walker" is specific and gets better results. AI responds to clarity.

Why Specificity Matters for Accessibility

When you're generating accessible content, you're often solving for multiple needs at once. Someone might need alt text that's descriptive enough for a screen reader but concise enough to fit technical constraints. Another person might need complex medical text simplified without losing accuracy. Generic prompts don't handle these competing requirements well.

A weak prompt: "Make this more accessible." What does that mean? Easier to read? Screen reader compatible? Shorter? The AI guesses.

A strong prompt: "Rewrite this medical explanation for someone with a high school reading level who uses a screen reader. Keep it under 200 words, avoid jargon, and if you must use technical terms, define them immediately after."

Building Better Accessibility Prompts

Start by being explicit about the end user's need. Are you writing for someone using a screen reader, someone with dyslexia, someone who's Deaf and needs captions, or someone with cognitive disabilities? Different disabilities benefit from different formatting and language choices.

Then specify constraints and format. Should the output be bullet points or paragraphs? How long should it be? Should it follow a particular structure? Should certain words be avoided?

Finally, ask for what you'll check. "Include markers [like this] where you've simplified complex terms so I can review them" helps you catch mistakes before they reach the actual user.

Real-World Prompt Structure

A solid accessibility prompt follows this pattern: (1) The task—what you want done. (2) The user—who will use this. (3) The constraints—format, length, complexity level. (4) The format—how you want it structured. (5) The review step—what to include so you can verify quality.

Example: "Convert this meeting transcript into accessible notes for someone with ADHD and auditory processing disorder. Break it into short sections with clear headers. Include timestamps for key decisions. Use simple sentences under 15 words. Highlight action items in bold."

Try this: Take a piece of content you want to make more accessible. Write two prompts for the same AI tool: one vague (like "make this accessible") and one detailed using the structure above. See how different the outputs are. The detailed version will be closer to what you actually need, saving you revision time.