Social and educational AI for diversity: research on democratic values to develop artificial intelligence tools that guarantee access for all to educational tools and public services

Innovations in Artificial Intelligence for social development must fulfill a dual objective: improving equal access to areas of public interest such as education, training and public services, as well as improving civic and democratic participation in the process of developing such innovations for all.
A particular case will be made concerning equal access for all people to public services, such as legal or educational procedures. This accessibility requires cognitive adaptability because often the legal or administrative language is very complex. Not only for people with cognitive disabilities, but also for older people or citizens at risk of educational or social discrimination. Natural language processing technologies using Artificial Intelligence can provide tools to translate legal, administrative or educational texts into a simpler language accessible to all. Despite technological advances in language processing and machine learning, this becomes a huge project if we really want to respect the ethical and legal consequences. This kind of respect can only be achieved with civil and democratic engagement in two areas: 1) democratically selecting the type of texts and the type of translations needed and 2) involving citizens, experts and non-experts, to produce and validate real examples of legal texts with cognitive adaptations to feed the artificial intelligence algorithms that will be used to learn how to translate those texts.
The social value of this model goes far beyond building a tool to translate texts into a simpler and more accessible language, adapted to any type of population. The model aims to define participation and validation criteria to avoid the legal, social and equity biases present in the design of generative AI language tools for social purposes.