Google’s AI in the real world helps 38 million farmers predict a tricky monsoon season

In a sea of news about AI-powered chatbots, image generators, and smart assistants, it’s easy to lose sight of the truly profound, world-changing potential of this technology. But today, Google has shared a story that is a powerful and inspiring reminder of what AI is capable of when applied to a real-world problem.

A new post on the Google Research blog details how one of its AI models, NeuralGCM, was a key part of a project that delivered life-changing, advanced monsoon forecasts to 38 million farmers in India this summer.

Making supercomputing accessible

For years, highly accurate weather and climate modeling has been a costly and complex endeavor, often requiring the immense power of a supercomputer to run. With their NeuralGCM model, Google Research set out to change that. By combining traditional physics-based modeling with machine learning trained on decades of historical weather data, they created a model that is not only highly accurate but also incredibly efficient.

The key breakthrough is that NeuralGCM can run on a single laptop, effectively democratizing access to high-quality forecasting and putting a once-unattainable technology into the hands of a much wider scientific community.

Solving a century-old challenge

After Google open-sourced the model, the University of Chicago’s Human-Centered Weather Forecasts Initiative put it to the ultimate test. They recognized that accurately forecasting the start of the monsoon season—a century-old challenge—is one of the most critical pieces of information for hundreds of millions of farmers whose livelihoods depend on it.

By blending NeuralGCM with other advanced models and historical data, the university team was able to accurately predict the onset of this year’s Indian monsoon up to a month in advance, even capturing an unusual dry spell in its progression.

Incredible real-world impact

In partnership with the Indian Ministry of Agriculture, these pioneering AI-powered forecasts were delivered via simple SMS text messages to 38 million farmers. This timely information allowed them to proactively adjust their planting decisions to adapt to what was an unusually delayed monsoon season.

The impact of this cannot be overstated. Existing research from the University of Chicago shows that providing farmers with an accurate forecast a month in advance can lead to an almost doubling of their annual income.

This is a truly superb and humbling use of AI. It’s the perfect example of taking an impossibly complex technology and making it accessible and useful for the people who need it most. It’s stories like this that make me genuinely excited for the future of AI. If we can do this for monsoon prediction, what’s next? It makes you wonder just how accurate and helpful our everyday weather forecasts might become in the very near future.