It took science about 40 to 50 years to reach the stage where weather could be predicted with around 80pc accuracy.

But in the past few years alone, AI has helped forecasters to achieve a 90pc rate of confidence for their predictions, up to 10 days in advance.

Now a new AI-led weather forecasting service for Ireland has been set up, promising better predictions of the intensity of future storm and flooding events.

The new system is also intended to result in far less “extreme event” ­surprises like Storm Éowyn in January which caused mass power outages.

The Aimsir (AI for Meteorological Services Innovation and Research) ­Centre has been launched with a €5m investment from Met Éireann, Ireland’s National Meteorological Service.

The goal of the centre is to transform weather forecasting and help Ireland better prepare for climate risks, and it is headed up by Andrew Parnell, the Met Éireann professor of data science for climate and weather at University College Dublin (UCD).

“For a long time, weather forecasting was all about solving complex mathematical equations,” Professor Parnell said.

“We’ve relied on computers for decades, incrementally building up accuracy.”

In the past three years, Prof Parnell said, AI has become more important for weather forecasting than solving equations.

“These tools, similar to those behind ChatGPT, can take in huge amounts of weather data, almost throw away the equations, and just learn from ­examples to forecast what’s coming next. It turns out they can be even more accurate than the traditional approach.”

The leap forward in forecasting accuracy provided by AI has been swift.

“This isn’t just more numbers,” Prof Parnell said.

“It’s a material improvement that’s already bringing better results for ­everyone from farmers planning the season to emergency responders preparing for storms.”

The Aimsir Centre will support 20 research projects – each one pairing an academic expert and senior Met Éireann forecaster – and 20 PhD students tasked with improving weather predictions.

The goal, Prof Parnell said, is not just to provide better rain and temperature forecasts, but to answer questions that really matter such as: When will the next storm hit? Will there be flooding or heat waves? Can we pinpoint ­power lines that might be felled by wind?

“This is about synthesising data; combining traditional forecasts with new information on everything from ­river levels to factors that drive extreme weather events,” Prof Parnell said.

“We are trying to forecast outcomes that matter deeply to the Irish public.”

Weather is complex and hard to predict, as it’s swayed by countless competing factors, and is prone to surprises.

Handling the data mountain needed to do it well requires huge computing power and that’s why UCD has just installed a new AI infrastructure platform.

This platform is a mix of all the hardware, software and networking systems needed to teach the AI how to better predict our weather.

“It was turned on recently; they say it can handle 16 quadrillion calculations per second,” Prof Parnell said.

“That’s the equivalent of every ­person on Earth solving a million maths problems in one second. And that’s just a small AI machine – not even a true ­super-computer – but it gives us the horsepower we need.”

Yet, even with AI backed up by such remarkable computing power, ­limitations on forecasting accuracy will remain.

“Extreme weather events, those big disruptive storms, remain incredible hard to pin down. We’ll never reach 100pc accuracy. Sometimes, yes, we will still be caught out.”

The stakes involved with weather prediction in Ireland are high, with farmers depending on accurate rainfall and drought forecasts to guide their planting and grazing decisions, while flood warnings help local authorities and emergency services with their responses.

Ireland’s security also relies on accurate forecasts, and as tensions rise globally, they are more important than ever to support our military’s efforts to patrol and secure Irish waters and airspace.

“Our forecasters are based in every Irish airport, and coastal predictions are vital,” Prof Parnell said. “Fog, in particular, is troublesome. Predicting when it will set in is very difficult but with ­airport weather stations and AI, we’re aiming to improve those local forecasts.”

The European Centre for Medium-Range Weather Forecasts (ECMWF) is integrating AI into its operational tools, and forecasters here use these tools to prepare RTÉ weather bulletins.

The expectation is AI-augmented weather predictions will become the ‘norm’, enabling much more detailed, localised reports.

But, will our forecasters ever be able to predict the abrupt, unpredictable rain showers? “We are hoping that by combining traditional methods with AI-driven analysis we’ll see real improvements for those tricky conditions,” Prof Parnell said.