Using AI Tools to Write Image Descriptions
Generative AI tools can be used to assist in the creation of image descriptions, but a human needs to understand the context and purpose of the image in order to evaluate the output. The output should also be reviewed to ensure it is accurate, not omitting important information, or adding unnecessary details.
AI Tools
- Image Description Specialist (custom ChatGPT created by WCLD)
Links to an external site.
- This tool was created by the WCLD AI team in consultation with the Accessibility Team to ensure it follows accessibility best practices to create alt text and long descriptions. Please note it requires a free account for limited use and a paid account for unlimited uses. For WCLD staff, more information about how to use this tool is available on the Image Description Specialist Documentation Links to an external site. page in WCLD 101.
- Arizona State University Image Accessibility Creator
Links to an external site.
- This tool is free and offers unlimited uses, but it does have a tendency to provide wordy descriptions that need to be refined for clarity.
- Perplexity
Links to an external site.
- This tool is not specifically for image description, but it can describe images.
- It draws on current information and reputable sources.
- Penn State users have unlimited use of this tool until October 2025.
What to pay attention to when using AI
- Edit descriptions to ensure they don't contain unnecessary details such as the color and placement of objects in complex graphics
- When asking it to transcribe text in an image, it may occasionally leave some of the words out.
- It may put quote marks in alt text, which need to be removed as they interfere with the HTML code.
- AI descriptions often include insights or editorializations, which should be excluded from an image description.
Benefits of Using AI
In addition to generating an image description, AI can help save time by doing the following:
- Convert an image of a table into an accessible HTML table
- Convert an image of math into MathML or LaTeX code
- Extract the text from an image
Updated May, 2025