Redes de Investigación 2024

Natural language is the most common and efficient tool for humans to convey information. People use language, our natural means of communication, to encode, store, transmit, share and manipulate information. Natural Language Processing (NLP), therefore, is one of the most important and disruptive functional application areas of Artificial Intelligence in the current digital transformation. Language Technology is currently undergoing a paradigm shift due to deep learning techniques and pre-trained models, also called Language Models (LMs) or Fundational models. These models have emerged as transversal tools that dominate this area of ​​AI. LMs are being trained by large technology corporations, which means that they do not meet the needs or specifications of speakers, institutions or companies of a language. In this sense, it is essential to advance local research to build LM, and this is how the Ministry for Digital Transformation and the Civil Service understands it by launching the PERTE for the new language technology, which has been followed by the ALIA initiative (Axis 1 of the SEDIA Artificial Intelligence Strategy 2024) and the autonomous plans Nos, GAITU and AINA that cover Galician, Basque and Catalan. These initiatives have a clearly finalistic objective, the construction of LM for the official languages. On the other hand, the technological mastery of LM comes from the hand of a few research and development centers in the world, both public and private, with a critical mass of researchers and experts of excellence, who are those who researched a few years ago in the techniques that have later led to the advances in LM. The centers that make up this proposal are involved in the above-mentioned government initiatives, and are regular interlocutors of the relevant positions of the Ministry of Digital Transformation and the Civil Service. The ultimate goal of the RutaMdL network is to articulate a coordinated basic research strategy so that the country's centers can compete with researchers in the area and focus on advancing the frontier of knowledge in LM. The network is therefore complementary to the aforementioned government initiatives, as it will focus on analyzing which areas of research in the LM are the priority and most promising and will draw up a roadmap of strategic interest that will allow it to advise political authorities and thus achieve scientific leadership in these areas, as well as study strategies to increase the impact of the research results obtained in these projects. This network will also promote collaboration, coordination and the creation of synergies and complementarity between the most relevant research groups in the LM area, all of them funded in previous calls for proposals from the State Plan for R&D&I and involved in the aforementioned government initiatives. The IP is also responsible for the Chair of Artificial Intelligence and Language Technology funded by SEDIA, unique in its area, and IP of the project proposal HumanAIze: Humanizing AI with language technology in the call for research plans in cooperation in artificial intelligence of the Ministry of Universities.