Trustworthy and reusable GenAI services for Earth System Sciences

In Earth System Science (ESS), Generative Artificial Intelligence (GenAI) is emerging as a transformative technology, supporting faster and more collaborative science by opening up new ways to discover and use data across domains. Earth System Sciences, with their deeply interconnected varied domains can particularly benefit from these new opportunities for the integrated use of heterogeneous, multi-disciplinary data sources (such as satellite and in-situ observations, models, experiments, or texts) hosted in diverse repositories managed by national and European infrastructures. GenAI services can help map, complete, convert, and process ESS metadata and data to enable cross-disciplinary studies and applications.

This is the core ambition of GenAI4Earth: to unlock generative AI’s potential by embedding it responsibly within the EOSC ecosystem and work closely with user communities and research infrastructures to co-design services that meet their needs.

GenAI4Earth will deliver explainable, benchmarked, and reusable services (metadata generation, dataset enrichment, foundation models), all machine-actionable, interoperable, FAIR (for data and models), and legally compliant.