MIT scientists have developed a training method meant to make climate emulators — AI models that quickly simulate how human activity can affect the future climate — faster to build and more adaptable. The team reports that an emulator trained on a smaller, carefully chosen set of synthetic data could still predict outcomes of scenarios it had not seen during training.
New research could lead to faster, trustworthy AI models for climate prediction
Posted by budbl.com (#1) 32 minutes ago (https://news.mit.edu)Who Voted for this Story
Copyright © 2026 budbl
| About
| Advanced Search
|
RSS
| RSS Feeds
RSS
| RSS Feeds