ERAI

Description

The ERAI project aims to promote a new generation of artificial intelligence technologies designed to adapt texts to Easy-to-Read (ETR) formats, significantly improving access to information for people with reading difficulties. Building on the advances achieved in the IRAZ project, ERAI addresses current limitations in data, models, and technological capabilities by incorporating new GAI architectures, autonomous agents, and advanced models in Basque to generate high-quality adaptations. In doing so, it seeks to optimize processes that are still largely manual today and offer more powerful, scalable, and accurate tools for the informational inclusion of all citizens.

Objective

ERAI’s central objective is to research and develop new Easy-to-Read adaptation systems based on generative AI technologies, autonomous agents, and advanced models for Basque and Spanish, enabling the generation of accessible content with less post-editing effort and higher quality. The project aims to raise the sector’s technology readiness level (TRL), increase the availability of suitable models, and consolidate an innovative solution capable of supporting organizations working toward information accessibility, meeting the field’s real needs and promoting social inclusion.

Actions

• Analyze the technological limitations identified in the IRAZ project.
• Investigate new high-capacity IAG models applicable to Easy Reading.
• Study the use of autonomous agents to generate, evaluate, and refine adaptations.
• Integrate IAG models specific to Basque with the ability to follow instructions.
• Develop advanced prototypes for automatic adaptation to Easy Reading.
• Evaluate improvements in the quality, variability, and accuracy of the generated adaptations.