Weather prediction tools powered by artificial intelligence fall short when forecasting record-breaking temperature and wind events, according to research from the University of Geneva.
What's happening?
Scientists analyzed global weather information from 1979 to 2020 to test how well artificial intelligence systems predict extreme conditions. They identified more than 162,000 heat records, nearly 33,000 cold records, and over 53,000 wind records in 2020.
When comparing AI tools against the European Centre for Medium-Range Weather Forecasts' physics-based model, the traditional system outperformed every AI option across most timeframes.
"AI-based models are revolutionizing weather forecasting and have surpassed leading numerical weather prediction systems on various benchmark tasks. However, their ability to extrapolate and reliably forecast unprecedented extreme events remains unclear," the researchers noted.
The AI tools often underestimated how hot heat waves would get and overstated how cold freezing events would become. These systems also failed to detect many extreme events altogether.
Why is AI weather forecasting concerning?
The problem comes down to how these tools learn.
TCD Picks » Upway Spotlight
💡Upway makes it easy to find discounts of up to 60% on premium e-bike brands
AI systems train on historical patterns, which means they struggle when conditions exceed anything they've previously learned from. Traditional models use equations rooted in physical laws, which allows them to model rare atmospheric conditions.
This gap poses the biggest problems when the stakes are high. Planning for dangerous heat, managing electrical grids, and preparing for storms all depend on accurate extreme weather warnings. Underestimating dangerous conditions puts communities at risk.
AI systems also carry broader environmental costs. Data centers powering these tools consume massive amounts of electricity and water for cooling. While AI can help optimize clean energy grids and improve efficiency in some areas, widespread adoption increases demand on power systems, potentially raising energy bills and straining resources.
What's being done about AI weather forecasting?
Researchers are exploring several fixes.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
Scientists can feed these tools synthetic weather scenarios generated by physics-based models. Doing so exposes the systems to extremes that actual historical data doesn't include.
|
How often do you use AI tools? Click your choice to see results and speak your mind. |
Every day 💯 A second path combines the quick processing of machine learning with the reliability of physics-based calculations.
If you want to stay prepared, sign up for local emergency alerts. Follow guidance from national weather services during extreme events.
Support policies that fund weather prediction research and climate adaptation programs. This strategy can also help communities build resilience against worsening extremes.
Get TCD's free newsletters for easy tips to save more, waste less, and make smarter choices — and earn up to $5,000 toward clean upgrades in TCD's exclusive Rewards Club.
Cool Picks
