The HumanAI project was presented by Paloma Martínez Fernández from the HULAT – UC3M (Human Language and Accessibility Technologies Group) at Universidad Carlos III de Madrid, during the event “Clear and Accessible Language and Artificial Intelligence”, held at the Royal Spanish Academy (Real Academia Española) on 26 and 27 May.
The presentation took place in the session devoted to cognitive accessibility and highlighted the importance of developing artificial intelligence systems that support clear language and Easy-to-Read content through a human-centred approach.
Within HumanAI, we are exploring the use of open language models, with particular attention to smaller, more sustainable models, to generate more accessible and personalised text. The project does not aim to replace human judgement with automatic generation. Instead, it promotes a controlled and assessable process in which users and experts participate at different stages.
Paloma Martínez emphasised that the quality of AI-generated simplifications cannot be evaluated only through automatic metrics. It is also necessary to verify whether the text preserves meaning, avoids removing relevant information, does not introduce invented content, respects sensitive terms, and fits the needs and profile of the target users.
This approach reflects the main goal of HumanAI: to advance towards artificial intelligence systems that are accessible, responsible, governed and centred on people.
The full presentation is available from 1:05:
https://www.youtube.com/live/1n8HzzkNAB8?si=ZU5eTrAnRX9ckIVl
