Generative AI: Simplifying Text for Cognitive Impairments and Non-native Speakers

It is well known that people with cognitive disabilities, older people, and immigrants face barriers in understanding technical texts. Some organisations makes efforts in developing guidelines for the creation of Easy-to-Read texts, although these have limitations, such as a lack of specificity and a focus on intertextual cohesion. There have been several recent papers that put forward new techniques for simplification like the addition of notes and Q&A sections while others emphasise a balance between simplification and quantity of information. They also criticise that conventional simplification methods are dependent on expertise and resources that may not be available and that there is a lack of focus on the individual needs and different target groups, so artificial intelligence presents an opportunity to expedite these processes. The results of the use case and the comparison between the resulting LLM’s yielded correct components but also shortcomings. These give room for further exploration to consider optimising prompts for better results.