A student who has spent years solving problems from other people's textbooks suddenly receives three notebooks with real ocean observations. There are few problems, but they have already seen hundreds of similar ones, and now a couple of examples suffice to catch the new teacher's style. This is how a diffusion model (a neural network that learns to remove noise from data) masters the climate of the Pacific Ocean, where the warm El Niño phase alternates with the cold La Niña and governs weather across the planet.

The trouble is that satellite and buoy measurements of sea surface temperature only began in the mid-20th century. Since then, about 720 monthly mean maps have accumulated, while neural networks with millions of parameters typically need thousands of examples. To honestly check what the model actually learned, the authors built a testbed: they generated 120,000 synthetic fields with known structure. On this data, the model accurately reproduced both symmetric and asymmetric variability, but as soon as the sample was reduced to the real size, it started memorizing training examples instead of generalizing.

The way out was suggested by the nature of learning itself. First, the model was pretrained on synthetic fields from 30 climate models of the CMIP6 project, adding an eight-dimensional vector to the architecture to encode each model's identity. Then, on the 720 real observations, only this vector was fine-tuned. This approach achieved pattern correlation above 0.85 and asymmetry correlation above 0.9, outperforming classical linear models. A student trained on many virtual worlds masters the one real world after just a few examples.

This changes the view of how generative models can work in Earth sciences: not as a replacement for physical models, but as a tool for extracting structure from sparse empirical samples. The method can be applied to generate realistic ensembles of El Niño states for climate risk assessment, and to improve statistical forecast models under data scarcity. The student who started with borrowed textbooks is now writing new pages themselves.