Easy Read Documents with AI

Easy Read developed alongside self-advocacy efforts by people with intellectual disabilities. Organizations such as Mencap and CHANGE publish Easy Read materials and guidance. Common practices include short sentences, one idea at a time, familiar words, clear layout, and pictures that help explain the text. Producing Easy Read by hand is slow and requires trained writers plus a symbol or photo library, which is why many organizations only convert their most-used documents.

AI language models can now draft an Easy Read pass by applying learned rules -- shortening sentences, replacing abstract or Latinate words with concrete ones, and flagging places where an image would help. Some tools pair this with image generation or stock symbol libraries to suggest illustrations automatically. Research has explored LLM-assisted Easy Read drafting, but results depend on language, document type, and review process; a faster draft is not evidence that readers understand it.

The common misconception is that Easy Read is just 'simple language' or a reading-level score. It is a distinct format with formatting conventions (large font, generous spacing, left-aligned text, one point per page section) that plain language alone does not capture, and it is built for people who may not read at all and rely partly on the images. AI-drafted Easy Read text still needs review by someone trained in the format, and ideally by a person with lived experience of intellectual disability, because AI can misjudge which words are actually simple for this audience (it may keep a word that sounds simple to a general reader but is unfamiliar in context) and can generate images that don't match the text precisely enough to support comprehension.