Today's Edition
ISIDORA CIRIC (ABU DHABI)
Farmers in poverty-stricken, remote areas often plant in hope and harvest in uncertainty. But a new AI-powered project is trying to change that by putting accurate, locally relevant weather forecasts into the hands of those who need them most.
The AIM for Scale Weather & AI-based Weather Forecasting project is part of a wider effort to fix a broken system - one where decisive weather information often doesn't reach smallholder farmers, and when it does, it arrives too late or too generic to help.
"We're trying to make weather finally work for people," Amir Jina, AIM for Scale Technical Panel Chair and Assistant Professor at the University of Chicago, told Aletihad in an interview.
The project forms part of a wider initiative called Infrascale, which identifies proven but underused innovations in agriculture and aims to scale them for real-world impact. In its first year, the focus has been on weather, and more precisely, on delivering forecasts that farmers can actually rely on.
This is where AI comes in. Traditional weather forecasting relies on physics-based models that run on multimillion-dollar supercomputers, Jina said. However, these systems are out of reach for many low- and middle-income countries, both in terms of cost and customisation.
"Only the richest countries could afford them," Jina said. "That meant that these systems focused on producing the best and most accurate forecasts for these countries' areas, because that's what they spent their money on."
But in the past three years, AI has upended that model. New tools, developed by teams including one of Jina's collaborators at Nvidia, have shown that AI models can not only match but outperform conventional forecasts, at a fraction of the cost and time.
"This completely democratises access. Instead of running on a supercomputer, you run this on a laptop," he said. "That means that every single country that could never afford to do this before can suddenly produce its own forecasts."
More than just offering access, the project is about precision. Instead of one-size-fits-all predictions, these AI models can be tailored to the decisions farmers need to make.
"What do our farmers need? Rainfall information right when they're planting their main crop? Then we make the model accurate for that decision," he said. "That's the difference between saving or losing a harvest."
This ability to customise forecasts down to the regional or even local level is a breakthrough. The models already outperform conventional ones on one-to-ten-day forecasts by as much as a full day - a gain that took traditional systems decades to achieve.
But AIM for Scale isn't stopping there. The team is investing in research to extend what they call the "lead time" of forecasts, pushing predictions further into the future and tailoring them to different agricultural decisions.
Around this innovation is a wider infrastructure push that includes improving weather observation networks, building tools to assess the usefulness of forecasts for farmers, and experimenting with how best to deliver that information - via phones, radio, or other media.
The scale of the ambition is matched by the investment behind it. The AIM for Scale weather programme was selected for support under a $200 million fund announced at COP28 in Dubai, jointly established by the UAE and the Bill and Melinda Gates Foundation. That support has since helped coordinate a broader $1 billion effort to modernise global forecasting for the poorest farmers.
Jina is clear that this backing has done more than unlock funding. It's changed how the work gets done.
"We wouldn't be here without this partnership," he said, adding that the people behind the fund are not just providing capital, they're working alongside researchers to remove the roadblocks that usually stall good ideas.
Scaling the project could unlock billions in benefits for farming communities. Improved forecasts in India alone could generate more than $3 billion in economic gains. In field trials in Benin and Colombia, the average benefit per farmer ranged from $103 to $356. And with climate change making weather patterns more unpredictable, those numbers are only likely to rise.
What it comes down to, Jina said, is helping the world's poorest farmers make better decisions in the face of an uncertain future.
"One wrong weather forecast can ruin a whole season. Now farmers in other parts of the world will actually have systems producing forecasts customised to their exact region's needs."
