The weather forecast sees the world in 25-kilometre squares. Your entire city fits inside one of them — along with the hill, the valley and the coastline that make your weather completely different from the next suburb. Jonathan Ferrer builds AI that fills in what's inside that square.

From a lab in Abu Dhabi, he takes blurry global forecasts and sharpens them to 200 metres: enough to say which streets turn dangerous in a heatwave, or whether a farmer needs to harvest this week. It runs on the same technology that generates images from a text prompt. It starts with static and refines until real weather appears. We talk about whether AI is worth its energy cost, whether this is a bubble, and why he insists that the old-fashioned meteorologists are now more necessary, not less.

Timestamps

• 01:30- From Spain to Australia to Abu Dhabi
• 03:04- Research at the University of Queensland and CSIRO
• 04:03- Why applied science matters for real world impact
• 09:29- Why generative AI matters for weather and climate
• 13:48- Working with governments on climate and infrastructure decisions
• 16:24- Balancing AI cost, energy use, and public value
• 20:57- Ethics, transparency, and energy efficient deployment
• 31:43- Agriculture, crop planning, and localised weather risk
• 35:04- Why traditional models and domain expertise are still essential

Follow us! Instagram: sciencebehindtheglass

X: @sbtg_podcast @daniesguerra @LINDAHE1988

FB: Science behind the Glass

TikTok: https://www.tiktok.com/@sbtgpodcast?_t=ZS-8u2DmABj4Us&_r=1

We have a new blog! * https://sciencebehindtheglass.com/blog