Global news view
¼ billion
URLs processed each day.
AI generated Human Centered Understanding of Extreme Weather Events
The system is an AI enabled Human Centred Evidential System to increase awareness, build trust, and enable targeted accountability at the granularity of local reporting with planetary scale. It helps change the architecture of the conversation by providing the textual evidentiary support that can feed into the “Loss and Damage” discussion, help build trust in understanding events for financial institutions to build innovative products, and empower the donor community to target their support much more locally.
By having these planetary scale maps we can build trust to see the momentum of what is happening when and where, and use a breakthrough AI capability to be a platform for breakthrough climate action.
Planetary scale event map
Key metrics
Translation
20
Languages translated for genuinely better coverage.
NLP pipeline
Planetary scale
Natural language processing: parsing, de-duplication, named entity recognition, classification.
Location API
⅓ million
Unique places in a knowledge graph, with over a million alternative names.
Human timelines
Human scale
Events extracted within human timelines rather than machine windows.
AI event matching
~180K
Reports detected per year, evaluated on certainty, severity and related measures.
About ClimateGPT
This map is one view onto the corpus behind ClimateGPT, an ensemble of open source large language models built to augment human decisions on the impact of climate change. The extreme weather events plotted here are part of what it reads.
Its corpus was drawn from over 100 billion web pages and millions of open-access academic articles, including research on climate, extreme weather, the Club of Rome’s Earth4All and the UN Sustainable Development Goals. The foundational model was trained and hosted entirely on renewable energy, and answers in more than 20 languages.
