WeatherNext 2 improves accuracy across most weather variables and forecast lead times, extending up to 15 days, compared with its predecessor.

However, experts caution about reliance on historical data in a changing climate, the need for transparency in AI predictions, equitable access, high training and deployment costs, and the risk of outlier events where data is sparse.

The article highlights practical benefits for everyday planning through clearer insights into what could happen and when.

The piece references broader context and authorship, noting Jonathan Kemper’s coverage of how AI tools can enhance work and creative projects.

WeatherNext 2, Google DeepMind’s AI weather model, runs on a Functional Generative Network to deliver forecasts eight times faster with more detail, enabling hundreds of potential scenarios in minutes.

The model is already embedded in Google Search, Gemini, Pixel Weather, and the Weather API, with plans to extend integration into Google Maps.

The discussion touches on potential cross-market effects, including AI-themed ETFs with crypto exposure and correlations between stock moves and crypto assets.

Energy traders could benefit from improved load forecasting, renewable integration, and risk management in day-ahead and hour-ahead markets.

The initiative aims to translate cutting-edge research into practical tools, expanding open data access and encouraging global scientific discovery.

No additional technical specifications or metrics are provided in the excerpt to quantify the claimed improvements.

Experts warn that shrinking public data infrastructure could undermine AI weather advances, stressing the need to protect open data sources while leveraging AI for public safety and research.

Attention is placed on practical use cases for traders and utilities, rather than a deep dive into the underlying algorithms.