Google has achieved a revolutionary first. The giant search engine has developed an algorithm that will discredit meteorologists. GraphCast, the search giant’s new artificial intelligence, creates efficient and safe weather forecast reports. Here are the details..
Google’s new artificial intelligence, Graphcast, proudly delivers the weather!
Google has developed a robot that will tell you which jacket you should wear when going out. Google DeepMind, the search giant’s artificial intelligence-centered brain trust, broke new ground. It announced a new weather forecast model that outperforms traditional systems by more than 90%. A machine learning model called GraphCast produces 10-day forecasts that are better, faster, and more energy efficient than the tools running your weather app today.
“We believe this is a turning point in weather forecasting,” Google researchers said in a study they shared.
So how does this algorithm work? Generally, this current model for forecasting is called “numerical weather prediction (NWP).” NWP calculates current weather conditions based on principles of fluid dynamics, thermodynamics and other atmospheric sciences. Additionally, this model simulates changes and creates huge models. This algorithm is quite complex. It’s also very expensive and requires tons of computing power.
Bad news from Google for unused Gmail accounts. Google gave three months for these accounts. So what will happen after three months?
Rather than running simulations of how molecules will fly and crash into each other, GraphCast puts more emphasis on historical data. Thus, he breaks tradition. In other words, this algorithm is a machine learning model that makes predictions based on what has happened in the past. There’s a lot of fancy computer science involved. However, in general, it has a much simpler structure in terms of the level and number of calculations it requires.
GraphCast starts with data about Earth’s current weather and the weather six hours ago. It then makes a prediction about what the weather will look like six hours from now. GraphCast then feeds these predictions back into the model. It performs the same calculation. Thus, it reveals longer-term predictions.
