Thirty-eight million farmers across India received forecasts this summer from the Indian Ministry of Agriculture and Farmers’ Welfare accurately predicting the start of the rainy season, known as the monsoon. The information, thanks to the power of artificial intelligence (AI), provided them with advance warnings up to four weeks ahead of the monsoon season. This could be the future for hundreds of millions of smallholder farmers across the tropics who have typically used past experience as their guide or waited until the rains struck to make important farming decisions like what to plant, how much and when.
Beyond India, many low- and middle-income countries have historically not had access to the type of accurate, real-time forecasting available in other parts of the world. As climate change makes the weather more unpredictable, this information is needed more than ever. Fortunately, a revolution in weather forecasting—largely driven by AI—has found ways to generate high-quality forecasts at a fraction of the cost and time. Even more groundbreaking, it’s opening the door to forecasts that can be tailored to the specific needs of stakeholders on the ground, such as farmers.
The Human-Centered Weather Forecasts Initiative at the University of Chicago Institute for Climate and Sustainable Growth is partnering with governments around the world to leverage this opportunity—bringing low- and middle-income countries forecasts that can not only run on desktops instead of supercomputers, but can specifically target what citizens need to know. The team of climate scientists, AI experts and economists then work with governments to disseminate forecast messages, test their effectiveness, refine them based on user feedback, and scale them up. In doing so, the Initiative provides communities with an essential tool for climate adaptation.
“Weather forecasts optimized for people’s specific needs have an incredible return on investment. The Human-Centered Weather Forecasts Initiative works with partners around the world to bring weather forecasting services to hundreds of millions of people,” says University of Chicago economist Michael Kremer, a 2019 Nobel laureate and co-director of the Initiative. Kremer chairs the global Innovation Commission on Climate Change, Food Security and Innovation. The Commission’s secretariat, based at the UChicago Development Innovation Lab, reviewed evidence on a range of innovations for evidence of impact and cost effectiveness and selected provision of AI weather forecasts among a small set of priority innovations.
The India monsoon forecasts that went out this summer were a program of the Indian Ministry of Agriculture and Farmers’ Welfare, who partnered with an international team of researchers led by the Human-Centered Weather Forecasts Initiative. It’s a model for the future.
“Our idea is to see how we can turn this into a generalizable tool—not just for farmers all over the world, but for other use cases,” says Amir Jina, a co-lead for the Initiative and an assistant professor at the Harris School of Public Policy. “AI is changing what’s possible in weather forecasting, allowing us to make forecasts that inform the decisions we’re making to live and adapt to climate change.”
The team begins by learning about what forecast information is most useful from the people who need it themselves, often working with governments and communities on the ground. They then set out to create forecasts tailored to those needs.
This is where Pedram Hassanzadeh, an Associate Professor of Geophysical Sciences at UChicago who co-leads the Initiative with Kremer and Jina, comes in as the AI and weather expert on the team. Hassanzadeh, with Caltech’s Anima Anandkumar and other collaborators, co-created NVIDIA’s FourCastNet, the model that kickstarted the AI revolution in weather forecasting in 2022. Hassanzadeh, and his team at UChicago’s AI for Climate initiative, a joint program of the Data Science Institute and Institute for Climate and Sustainable Growth, is charged with developing and benchmarking forecast systems to find out what forecasts best fit the societal needs.
“The goal of this work is not only to develop better AI models and provide more people around the world with wider access to more accurate and tailored weather information, but also to strengthen global forecasting systems broadly, enabling organizations that disseminate forecasts to generate, evaluate, and improve their own forecasts,” Hassanzadeh says. “This will be most impactful through interdisciplinary collaborations, like this Initiative, that can bring the rapidly advancing theoretical work of AI and climate scientists out of the lab and out into the real world.”
AI models are not only more accurate but also skip the need for supercomputers for forecast generation, according to Mayank Gupta, a researcher at the Human-Centered Weather Forecasts Initiative. He says, “AI offers a huge opportunity for technological leapfrogging in forecasting, proving already to be more accurate for many types of weather forecast, and able to be easily expanded to everyone.”
The Human-Centered Weather Forecasts Initiative goes beyond working with governments to generate and select novel AI-based forecasts. It develops the local partnerships needed to disseminate the forecasts to citizens, and then tests how effective the new information is in delivering actionable change.
“We’re combining tailored weather forecasts with strategic partnerships and a user-focused approach that is often missing in forecast science,” says Jina. “In not only providing better forecasts but refining them based on user feedback, we’re working with governments to provide citizens with actionable climate information and ensuring they have the tools to adapt and thrive in an increasingly uncertain environment.”
The Indian government’s approach—grounded in the needs of farmers and collaborating with the Human-Centered Weather Forecasts team and other researchers to take cutting-edge science out into the real world—offers a compelling blueprint for the future. The team is working with AIM for Scale, a global initiative backed by the Gates Foundation and the United Arab Emirates, to scale similar programs in other low- and middle-income countries.
But providing farmers with critical task-specific information is just the beginning. For example, the Initiative is exploring how tailored forecasts of heat, humidity, night-time temperature and other variables could help those without access to air conditioning protect themselves—such as by learning the symptoms of heat stroke, arranging for adequate hydration, and protecting vulnerable populations like the elderly. The team will utilize the climate impacts data collected by the Climate Impact Lab—another initiative based at the UChicago Institute for Climate and Sustainable Growth—to target the regions and industries where the dissemination of tailored weather forecasts could be most beneficial, and learn further from the people impacted in those locations.
“There are many reasons why society may distrust AI, but here is a positive way we can make a revolution in technology really work to improve people’s lives throughout the world,” says Jina.
The Initiative’s work builds off the University of Chicago’s long history of pioneering economic thinking to tackle major societal challenges, as well as the less discussed advances in weather forecasting and climate modeling. In 1940, the meteorologist Carl-Gustaf Rossby came to the University to lead its newly established Institute of Meteorology, which eventually became the Department of Geophysical Sciences. While at UChicago, Rossby advanced revolutionary research in weather forecasting, which included identifying the importance of the jet stream in creating weather patterns in the high atmosphere.
Rossby and his students laid the foundation for what is known today as numerical weather and climate prediction, and trained cohorts of professional meteorologists in these cutting-edge efforts on the eve of World War II. Following suit, the Initiative will train forecasters from low- and middle-income countries on how to build and use AI weather forecasting models.
“Eighty-five years later, this Initiative builds off this work—ensuring it no longer remains in just the wealthier countries who can afford the infrastructure,” says Amir Jina. “We’re working to push the frontier of weather forecasting, focusing on the needs of those who could most benefit and helping to build the tools to get them there.”
The Human-Centered Forecasts Initiative is a University of Chicago Institute for Climate and Sustainable Growth Venture Fund project.