A public dashboard observing signals, trends and issues.
SubscribeLogin한국어
Latest observation
2026-10-08
Public objects
4434
Build time
2026-10-08 19:44 KST
The Futures

Signal GraphCast predicted a 2023 hurricane landfall nine days ahead, three days earlier than standard models

Summary

This signal shows AI outperforming conventional weather models on a major storm. In 2023, Google's GraphCast model correctly predicted where Hurricane Lee would come ashore in Nova Scotia nine days ahead, three days sooner than standard models. It is also estimated to be roughly 1,000 times more energy-efficient than traditional forecasting methods, which rely on supercomputers and expert interpretation. Other uses include an AI forecasting system in the Philippines for disaster preparedness. The report suggests AI could make forecasting cheaper and more widely available, narrowing the accuracy gap between rich and poor countries. It cautions, however, that AI models are less transparent and may miss unusual events not seen in past data.

Classification

Region menusGlobal
Occurrencescope:country · geo_region:north_america · country:CA
Impactscope:global
Time horizon0-3 years (2026-07-26)
Published2025
Last updated2026-09-30 12:56 KST

Evidence 1

Part of trends 1

Directly linked issues 0

No objects.

Relation types: supports

Public id: fm-04479ffeb693