MIT researchers have made a breakthrough in climate prediction by developing a technique to create fast, trustworthy, and adaptable artificial intelligence models. These climate emulators can simulate the effects of human activities on future climates, predicting impacts on temperatures or sea level. The new technique streamlines the production of these emulators, allowing scientists to efficiently test thousands of varied climate scenarios in a matter of seconds. This provides more reliable estimates and can identify unexpected outcomes of regional interventions. The researchers used a training technique that boosts the performance and adaptability of the climate emulator while using only a fraction of the data typically required. This significantly cuts production time for these practical scientific tools. The technique was tested with synthetic data, and the results showed that the emulator accurately predicted outcomes of scenarios it did not see during training, while traditional emulators failed to adapt to new situations. This breakthrough has significant implications for policy decisions, as climate emulators are well-suited for modeling policy-relevant scenarios. The researchers' method can speed up the production of more reliable emulators, enabling policymakers to make more informed decisions. The study was led by Christoher Womack, a postdoc at ETH Zurich, and co-authored by Shahine Bouabid, Andrei Sokolov, Popat Salunke, Glenn Flierl, and Sebastian D.