Humanizing AI with language technology (HumanAIze)

The "Humanizing AI with Language Technology (HumanAIze)" project seeks to enhance Large Language Models (LLMs) to make them more human, reliable, and ethical by addressing weaknesses such as bias, cultural understanding, and reliability. Six research centers bring their expertise in areas such as multilingualism, anonymization, verification, and evaluation. A key approach is "reliable by design," implementing ethical and legal principles, data governance, human feedback, and sustainability. HumanAIze will develop models with a focus on efficiency and extend to new languages and domains, integrating voice and visual modalities. The project will aim for openness and reproducible research, with an impact on the production of more capable language technologies and their use in practical applications, working to establish Spain as a leader in AI research and innovation.

Deskribapena (en):

The "Humanizing AI with Language Technology (HumanAIze)" project seeks to enhance Large Language Models (LLMs) to make them more human, reliable, and ethical by addressing weaknesses such as bias, cultural understanding, and reliability. Six research centers bring their expertise in areas such as multilingualism, anonymization, verification, and evaluation. A key approach is "reliable by design," implementing ethical and legal principles, data governance, human feedback, and sustainability. HumanAIze will develop models with a focus on efficiency and extend to new languages and domains, integrating voice and visual modalities. The project will aim for openness and reproducible research, with an impact on the production of more capable language technologies and their use in practical applications, working to establish Spain as a leader in AI research and innovation.

Deskribapena (es):

El proyecto "Humanizando la IA con Tecnología del Lenguaje (HumanAIze)" busca mejorar los Grandes Modelos de Lenguaje (LLMs) para hacerlos más humanos, fiables y éticos, abordando debilidades como el sesgo, la comprensión cultural y la fiabilidad. Seis centros de investigación aportan su expertise en áreas como multilingüismo, anonimización, verificación y evaluación. Un enfoque clave es "confiable por diseño", implementando principios éticos y legales, gobernanza de datos, feedback humano y sostenibilidad. HumanAIze desarrollará modelos con enfoque en eficiencia y extenderá a nuevos idiomas y dominios, integrando modalidades de voz y visuales. El proyecto apuntará a la apertura y la investigación reproducible, con un impacto en la producción de tecnologías lingüísticas más capaces y su uso en aplicaciones prácticas, trabajando para establecer a España como líder en investigación e innovación en IA.

Kode ofiziala:

AIA2025-163322-C61

Ikertzaile nagusia:

Eneko Agirre, Aitor Soroa

Erakundea:

Hitz

Hasiera data:

2025/12/03

Bukaera data:

2025/12/03

Taldeko ikertzaile nagusia:

Eneko Agirre