Could AI cost me my job as a weather forecaster?

Derek Brockway has been a weather presenter for the BBC for nearly three decades. He explains how weather forecasts are a vital part of our daily lives, helping us decide what to wear, whether to carry an umbrella, and when to hang out the washing. For industries such as farming, shipping, aviation, and renewable energy, accurate weather predictions are essential to their operations. Over the years, Derek has witnessed the increasing impact of climate change, which has led to more extreme and intense weather events. This makes improving weather forecasting more important than ever. Scientists are now exploring how artificial intelligence, or AI, could revolutionize the way we predict the weather by making forecasts faster, more accurate, and more efficient. Some broadcasters have even started experimenting with AI-generated weather presenters, raising questions about the future role of human forecasters like Derek.

Traditionally, weather forecasting depends on complex numerical weather prediction models that require vast amounts of data and powerful supercomputers, such as those used by the UK’s Met Office. Recently, the Met Office has partnered with the Alan Turing Institute, the UK’s national center for data science and AI, to develop a new global forecasting system powered by AI. One of their AI models, called FastNet, uses machine learning to enhance prediction capabilities. Professor Kirstine Dale, Chief AI Officer at the Met Office, describes AI as phenomenally fast—tens of thousands of times faster than traditional methods. This speed allows AI to produce up-to-date forecasts at a fraction of the computational cost and carbon emissions. AI can also generate hyper-localized forecasts tailored to specific postcodes, providing more relevant weather information to individuals. These AI-driven forecasts could improve early warnings for severe weather events like storms, floods, and heatwaves, helping to reduce their impact on communities. However, challenges remain, especially in predicting rare or extreme weather events. Because the climate is changing, past weather data is no longer a reliable guide for the future. Therefore, traditional numerical weather prediction models remain crucial for exploring climate changes and generating recalibrated datasets that can train AI models. Since AI models do not inherently understand the physics of the atmosphere, numerical models will continue to play a vital role in forecasting extreme events and validating AI outputs.

Dr Scott Hosking, Mission Director for Environmental Forecasting at the Turing Institute, highlights that once AI models are trained, they are cheaper and quicker to run than traditional forecasting systems. AI has shown surprising skill in predicting the paths of cyclones and hurricanes, learning from past data. However, it still struggles with forecasting high-intensity rainfall that can cause flash floods. AI also holds promise in space weather forecasting, which involves predicting solar storms that create phenomena like the aurora borealis. These solar storms can disrupt Earth's magnetic field and affect communication systems. Dr Huw Morgan, Head of Solar System Physics at Aberystwyth University, explains that space weather is complex and difficult to model because we cannot place recording stations on the Sun or between the Sun and Earth. Instead, scientists rely on remote data from telescopes. AI offers a valuable solution by helping to analyze this limited data and improve forecasts. Despite its potential, AI faces challenges in space weather forecasting, so traditional methods remain important. Regarding AI-generated weather presenters, Met Office meteorologist Aidan McGivern believes people prefer real presenters they trust over AI versions. He notes that weather forecasting has improved significantly over the past 18 years, with outlooks now available up to 14 days ahead. With AI, the future may hold even longer-range forecasts and more engaging visualizations for the public. Derek Brockway concludes that AI will likely complement rather than replace traditional forecasting methods, and for now, his job as a weather presenter remains secure.