AI and weather tracking as a very positive intervention

India’s monsoon season was unusual this year, but many farmers there had new AI weather-forecasting tools to help them ride out the storms.

Google’s open-source artificial intelligence model NeuralGCM and the European Center for Medium-Range Weather Forecasts’s AI systems are making sophisticated and granular forecasting data available to even the smallest farms in poor areas. Thanks to the open-source AI, and decades of rainfall data, the Indian government sent out forecasts to 38 million farmers to warn them about looming monsoons.

The initiative to help farmers adapt is the latest example of how companies are expanding their weather-tracking capabilities amid mounting concerns about extreme weather and climate change.

The effort is part of a growing “democratization of weather forecasting,” said Pedram Hassanzadeh, a researcher at the University of Chicago who focuses on machine learning and extreme weather. Researchers from the university partnered with the Indian government to gather and send out the monsoon predictions.

“Up until very recently, to run a weather model, you needed a 100 million-dollar supercomputer,” said Olivia Graham, a product manager at Google Research. But now, farmers in India can make better-informed agricultural decisions quickly, she said.

These projects seem to have very high benefit to cost ratios.  Here is one relevant RCT, here is another.  Here is more from the WSJ, via Michael Kremer.  Here is a useful and informative press release.