FUILD-AI

Abstract

The FAIR Principles have significantly advanced data management across research communities, yet they remain insufficient in the context of Artificial Intelligence, both at the data and model levels. As research infrastructures increasingly depend on AI and machine learning, the lack of interoperability and reusability of AI assets creates real barriers to scientific progress. FLUID-AI introduces the concept of Data and Models Liquidity, the idea that research assets — much like financial liquidity — should be readily available, interoperable, and easy to reuse across infrastructures. This concept extends the FAIR Principles into the AI domain and positions the European Open Science Cloud (EOSC) as a more dynamic, collaborative, and AI-ready research ecosystem.

The project pursues this vision through three pillars: building a community of practice and competence centre for training and support, achieving semantic and technical interoperability across data and AI/ML models, and developing an intuitive platform that lowers barriers for researchers regardless of technical background. The consortium, coordinated by IFCA/CSIC (Spain), brings together INRIA, KTH, EGI, KIT, UPV, EURO-BIOIMAGING, and MARIS, spanning infrastructure providers, domain RIs, and research universities. UvA contributes deep expertise in data management, semantic interoperability, and energy-efficient AI systems, leading work on state-of-the-art analysis, architecture definition, AI model interoperability metrics, and the Data and Models Liquidity baseline and verification criteria.