Generative Artificial Intelligence for Water

Generative AI (Gen AI) systems are general purpose Artificial Intelligences capable across a wide range of tasks characterized by the ability to analyse and generate text and images. Gen AI is highly effective at information analysis and synthesis and can operate at massive scale and speed. Gen AI technology is primed for revolutionising Environmental and Water Sciences, for example Gen AI can be readily augmented to perform geospatial processing and it can orchestrate structure and curate large sets of foundational environmental knowledge and data. 

Environmental Informatics pioneers, and delivers, Gen AI adapted to Water and Environmental Science. We’ve built the Roper River AI tool (RRAI) an advanced Environmental RAG system providing chat across over 3000 words of Environmental Reports. The RRAI is grounded in trusted science and we’ve developed a next generation citation mechanisms to anchor AI responses directly Water Science. The RRAI also incorporates dynamic mapping illustrating important locations across the Roper River on an interactive map responsive to user questions and conversations. 

The CSIRO Atlas of Water Science places our rich science library in the world by illuminating the unseen but critical connection of water publications to locations. Here we’ve analysed over 300 water science publications and mapped the relationship of the science described in the publication to places. This allows a quick visual understand of the proximity of diverse science revealing new patterns in knowledge. 

The Self Thinking Data Manifest proposes and demonstrates enhancing scientific publications with Gen AI. By including instructions crafted especially for Large Language Models (LLM) we show how publications can come to life providing users with rich interactive experiences powered by an LLM. 

We continue to innovate researching Gen AI for provenance generation, spatial data integration, environmental surveillance, trusted science, and environmental knowledge systems. We’d love to chat to you more.