July 12, 2024
In the rapidly advancing fields of Earth Observation and Earth Sciences, navigating and making sense of vast and complex data can be overwhelming. Non-specialists often struggle with intricate EO and ES concepts, policymakers require clear technical information, and researchers need more efficient tools to manage the ever-growing scientific literature.
We introduce EVE, an ambitious initiative funded by ESA. Our goal is to develop a highly specialised Large Language Model, fine-tuned as a virtual expert in Earth Observation. Eve will be capable of interpreting complex EO data and simplifying technical concepts, making them accessible to users of all backgrounds, from researchers and policymakers to the general public.
From now until summer 2025, EVE aims to revolutionise how we interact with EO and ES knowledge. By offering advanced capabilities in question answering, scientific summarisation, and technical support, Eve will enhance the way EO scientific material is produced and shared. The project will also provide open access to domain-specific data, trained models, source code, and comprehensive training recipes, empowering both current and future EO initiatives.
This groundbreaking project is led by Nicolas Longépé at Φ-lab (ESA's Earth Observation innovation lab) and executed by a consortium led by Pi School, in collaboration with Imperative Space. Stay tuned for updates on our journey to enhance Earth Observation with the power of AI!