Google's DeepMind team, which created GraphCast, believes the AI model can improve forecast accuracy by identifying patterns in the data. While traditional models are better at predicting events with multiple parameters, the creators of GraphCast aim to complement and assist these models rather than replace them. Google's CEO Sundar Pichai has announced a new team to further develop its AI technology.
Key takeaways:
- Google's AI model, "GraphCast," can predict the weather faster and with more accuracy than the standard European model, according to a published study.
- GraphCast uses historical weather data for training and can provide weather predictions in seconds, compared to the European model which can take over an hour.
- Despite its speed and accuracy, some researchers suggest that the early results do not guarantee the reliability of an AI-based model.
- Google's DeepMind team, which created GraphCast, was formed to further the company's AI technology and will power many of Google's future AI-enhanced products.