Alt text is a text description of an image that screen readers read aloud, making visual content accessible to blind users—but writing good alt text requires understanding context and purpose, not just describing what you see. AI can generate these automatically, which eliminates the blank alt text problem entirely, though the descriptions often need human refinement to be truly useful.
Alt text—short for "alternative text"—is a hidden description of an image. When a screen reader (technology that reads screen content aloud) encounters an image, it reads the alt text instead. Without it, people who are blind or have low vision simply hear "image" with no context about what's actually there.
Writing good alt text is an art. It needs to be descriptive enough to paint a picture, but concise enough not to overwhelm someone listening. A chart needs to explain its data. A photo needs to capture mood and context. A button needs to describe its function. Getting this right manually takes time and requires understanding both the image and accessibility best practices.
AI image recognition models can "see" images and generate descriptions automatically. You upload an image, the AI analyzes it using computer vision (a type of AI trained to interpret visual information), and it produces alt text in seconds. For a basic product photo or a simple chart, this works brilliantly. The AI might write: "Red coffee mug on a wooden table next to a notebook" in one click instead of you typing it manually.
But AI can make mistakes. It might miss important details (like text in an image), misinterpret what it's seeing, or write descriptions that are too generic. A chart showing housing prices over time might be described as "a line graph" without actually explaining the data. An image with cultural or emotional significance might get a surface-level description that misses the point entirely.
The real value is in speeding up the process and handling volume. If you're uploading 50 product photos to a website, AI can generate a first draft of alt text for all of them in minutes. You then review and refine—which is fast. Or for documents with dozens of screenshots, AI removes the barrier of "this is too tedious to describe manually."
The best approach combines AI with human review. AI generates the raw description, then you tweak it for accuracy and tone. This is faster than writing from scratch and ensures accessibility without creating extra friction.
Try this: Find an image on your computer or phone. Use an AI tool like Claude or Gemini to describe it, then ask it to write that description as alt text. Copy the alt text into an image editor or document. Then try reading it aloud or sharing it with someone—does it actually help them understand the image the way you intended?
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.