Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Subjects: | |
| Online Access: | https://doi.org/10.5281/zenodo.19686896 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- <p><span>We present ENVAI (Environmental AI), a network of nine specialised AI agents, each</span><br><span>representing a distinct European ecosystem—estuarine, riverine, lacustrine, marine, and for-</span><br><span>est.</span> <span>Each agent maintains a knowledge graph of sensor data, ecological events, species,</span><br><span>environmental drivers, and policy frameworks specific to its ecosystem. By constructing a</span><br><span>meta-graph layer (Numina) that maps shared concepts across the nine agent-level graphs, we</span><br><span>demonstrate the emergence of ecological knowledge that exists in no single agent’s dataset.</span><br><span>We identify eight shared species appearing across multiple systems, six continental-scale envi-</span><br><span>ronmental drivers, five cross-system climate events, four invasion pathways traceable across</span><br><span>national borders, and seven recovery timescales spanning four orders of magnitude from</span><br><span>days to millennia. We argue that a network of domain-specific AI agents, connected through</span><br><span>a shared ontological layer, can produce ecological insights that no individual monitoring</span><br><span>programme, institution, or country can generate alone. We term this capacity</span> <span>emergent eco-</span><br><span>logical intelligence</span> <span>and discuss its implications for cross-border environmental governance,</span><br><span>early warning systems, and the mapping of continental-scale knowledge gaps.</span></p>