The EU has developed in recent years a powerful set of instruments to unlock the full potential of data, including data platforms and data spaces; bring digital innovation closer to smaller actors and regional ecosystems through European Digital Innovation Hubs; foster AI-driven experimentation through AI Testing and Experimentation Facilities; and create ecosystems combining HPC, access to data, services and talent to develop and train large AI models through AI Factories.
This direction advocates for a clear pathway towards integrating these ecosystems into an integrated environment where data management and AI development complement and reinforce each other, enabling faster innovation, experimentation and deployment. This direction is based on the following priorities:
- AI-ready Data Products, a term coined by BDVA that extends the concept of Data Products to address the specific needs of AI and ML practitioners. This paradigm brings together data-sharing and data-value perspectives from data spaces with the requirements of AI/ML workflows. The concept should be further validated across industry sectors, supported through standardisation, integrated into data spaces, and complemented by an evolved framework for assessing the AI-readiness of data products.
- Data Labs emerging as a key connection between data and AI ecosystems by providing services to pool, curate, transform and prepare data from multiple sources for use by AI ecosystems, including AI Factories. Further efforts are needed to operationalise these instruments and establish a seamless connection between data infrastructures and AI development environments..
- AI-powered data spaces, integrating AI capabilities, including AI agents, to enhance data-space functionalities and services. Data spaces should be able to support AI-oriented data preparation and exploitation, while considering AI-driven data services and AI-enabled use cases as integral components of their value creation and business models.
- Strong integration between data infrastructures, AI platforms, experimentation facilities, and industrial deployment environments