Rural

WeatherWatch CEO praises AI for contribution to 'most accurate forecast data in the world'

The AI's ability to learn from its own mistakes is what Phil Duncan said sets it apart from other forecast data.

IBM forecast data and AI that's used by Weather Watch and Rural Weather in New Zealand has been deemed once again the world's most accurate weather data.

Weather Watch Chief Executive Phil Duncan told REX host Dominic George the IBM data is unique in its process of collaborating artificial intelligence alongside industry experts to create the most accurate weather forecast across the globe.

"You put humans in there, you put AI in there, you start tweaking that data and you start learning from your mistakes," Duncan said.

"That is the critical part about why our data with IBM is the most accurate in the world, it learns from its mistakes."

While there has been a consistent level of resistance to the continued advancements in AI technology in recent years, he sees it as an essential tool in accurate weather forecasting for more than just the country's most populated cities.

"Everyone is scared of AI, we're not.

"For weather, it is the best thing in the world because it gets into the local areas, not necessarily in all the main centres.

"We want those places in the middle of nowhere to be just as accurate as the main centres."

The IBM AI systems are able to analyse data and remember patterns and outcomes and apply those lessons to their future forecasts, a feature that sees their data stand apart from other providers due to its ability to learn from mistakes and adjust accordingly.

As an example of how this works, Duncan said if there were thunderstorms forecasted in Keri Keri which didn't happen. The AI system questions why they didn't happen, identifies where they actually did happen and remembers that the next time similar weather patterns occur.

While it is impressive, he said independent human analysis is definitely still required to create the most accurate weather forecast's possible.

"You need both AI and a human together. That's what we've got at IBM, WeatherWatch and Rural Weather and the more we work on that, the fewer false alarms we will get over time."

Duncan said despite their data being the most accurate WeatherWatch and Rural Weather still receive a significant amount of complaints from farmers in particular.

He understands that it can be frustrating when the weather can have such a significant impact on your work, but reiterated the fact that weather forecasts aren't 100% accurate, 100% of the time.

Listen to the full chat between WeatherWatch Chief Executive Phil Duncan and Dominic George above.

To check out more episodes listen to the REX Today Podcast anytime on your favourite streaming platforms including Spotify, Apple Music and Rova or tune in to REX Today live on Magic from 5 am to 6 am every weekday.

Visit the Rural Exchange Facebook or Instagram page for more information and regular updates from the REX team.