The developments and advancements in the space of AI take us by storm daily. Every day some new development comes across that we were not expecting at all. Today I am going to talk about Google’s AI tool – SEED. It is a generative AI model that predicts weather. Let’s find out everything about this new AI model.
Google’s AI tool SEED, which is short for Scalable Ensemble Envelope Diffusion Sampler, is being seen as a much-needed development in the space of weather forecasting. It is being said that this can help achieve the precision and reliability of weather predictions. It will help experts deal with atmospheric dynamics and uncertainty in quantification.
Also read: Google AI Boss: “Artificial Intelligence will be as intelligent as humans by 2028”
SEED moves over traditional weather forecasting methods by generating a huge database of forecasts to identify and characterise rare and extreme weather conditions accurately. This is different from the traditional methods that require substantial computational resources. It is expected to be a more smoother and cost-effective model.
As I said previously, Google’s AI SEED is exceptional because it can quantify uncertainty within weather forecasts, something that has always been an issue. Why am I saying this? So in meteorology, even minor alterations can lead to significant variations. This is called the ‘butterfly effect’. SEED’s accuracy will help with better planning for extreme weather occurrences.
If we talk about SEED’s computational efficacy it can give throughput of 256 ensemble members every 3 minutes on Google Cloud TPUv3-32 instances. This then helps create extensive ensembles which will help study rare yet impactful weather phenomena.
SEED was used during the 2022 European heatwaves. SEED’s predictions showed spatial patterns similar to those in operational forecasts. It also outperformed Gaussian models by accurately capturing correlations between different fields. It is expected that SEED will also help deal with the challenges of climate change. However, some wrinkles still need to be ironed out before Google’s AI tool, SEED can be used widely.
