Body
Alternative text, sometimes called an image description, is a textual substitute for images in web pages and digital documents. Artificial Intelligence has made generating these descriptions faster and easier, even for complex images like charts or infographics. This article guides you through using specialized university tools and general AI to create high-quality, concise image descriptions.
Always review AI-generated alt text for accuracy and the correct level of detail.
Binghamton University has its own custom-built AI Image Accessibility tool.
What it does: Generates alt text, detailed long descriptions, and extracts text from images or slides. It also provides LaTeX/MathML for equations contained in images.
How to use it:
- Visit hugo.binghamton.edu. Note: you must be on campus or on the VPN to use this tool.
- Paste an your image in the box using either Ctrl+V or Cmd+.
- Choose one of the options to receive your results.
Arizona State University's Image Accessibility Creator
The ASU Online AI Labs has developed a specialized tool for alternative text and image descriptions.
What it does: Generates alt text, detailed long descriptions, and extracts text from images or slides.
Privacy: This tool uses a secure API; your data is not stored or used to train external AI models.
How to use it:
- Visit the ASU Image Accessibility Creator.
- Upload your image (JPG, PNG, or JPEG).
- Optional: Use the "Add Details" field to provide context (e.g., "This image is for a Biology 101 quiz about cell structures").
- Click "Create Image Details" to receive your results.
Tips for Better AI Prompts
Standard AI tools (like Google Gemini, Microsoft Copilot or ChatGPT Edu) can sometimes be "chatty," providing more visual detail than needed. Use these prompting strategies to get concise, functional alt text.
Instead of asking "What is in this image?", ask the AI to act as an accessibility expert:
You are an expert in web accessibility. Write a concise alt text for this image (under 250 characters). Focus on the functional meaning and the 'why' of the image, not every visual detail. Avoid starting with 'Image of' or 'Photo of'.
Prompting for Context
Context changes what the alt text should be. With a precise prompt, you can get an image description tailored to your exact needs. Here are some examples that you can adapt to your own images:
For a data chart:
Describe the primary trend shown in this graph for a student who cannot see it. Include the key data points but keep it under three sentences.
For a photo that sets a tone:
This image is for a newsletter header. Provide a 1-sentence description that captures the mood/vibe.
For an Art History lesson:
This is [name of piece] by [name of artist]. Analyze this artwork in the context of its specific time period and movement. Describe the composition, color palette, and brushwork, and explain how these elements reflect the artistic goals of the era (e.g., Baroque's use of chiaroscuro). Keep the description objective but thorough for a student study guide.
For a STEM Diagram:
Summarize the process or system shown in this diagram. Focus on the relationship between components and the sequence of events rather than the shapes of the boxes or the direction of the arrows. Explain the 'scientific meaning' of the image in 3–4 sentences.
General prompt for alt text you can adapt
Using the provided image, create a description and alt-text. The image description should be objective, concise, and descriptive. Descriptions should be straight forward and factual, avoiding interpretations. Begin with a general overview of what the image portrays before providing details. Not everything needs to be described if it is not contextually important. Descriptions should utilize vivid terminology to describe various features like composition, shapes, size, texture, and color. Avoid using picture of, image of, and photo of unless needed in the context of an illustration or painting type graphic. The image alt-text should be concise, a few sentences at most. Should the image or graphic contain text, add a section and transcribe all the text presented.
Extracting "Flattened" Text from Screenshots
One of the most powerful uses of AI for accessibility is converting flattened text (text that is part of an image) into editable, digital text.
Why it matters
Screenshots of flyers, social media posts, or old scanned PDFs are "dead" to screen readers. AI uses Optical Character Recognition (OCR) to "lift" that text out. The Binghamton and ASU alt text tools highlighted above will do this for you. But you can also use general AI tools to extract text.
Upload the screenshot and use this prompt:
Transcribe all text found in this image exactly as written. Format it into a clean list/paragraphs so it can be used for a screen-reader-friendly document.
For Canva Users
In Canva (Magic Studio), the "Grab Text" tool allows you to select an image and instantly turn its flattened text into editable text boxes.