DeepMind beats classical baselines in scientific prediction with a new deep learning system.
DeepMind has published results showing how deep learning models can anticipate patterns in scientific data at an unprecedented scale.
The system was trained on time series from physics, biology, and climate, and outperformed classical baselines on several prediction tasks. External researchers have already begun reproducing parts of the experiments.
Experts warn that governance of these models will be critical to prevent misuse in sensitive domains.
On September 8, 2026 DeepMind launched AlphaGenome Atlas: ~9 billion SNV predictions, an AVI score, a ~1 PB-scale dataset, and academic access with GREGoR and UK Biobank collaborators.
Around September 4, 2026, DeepMind unveiled WeatherNext 3 with hourly forecasts on a 5 km grid (vs 25 km in WeatherNext 2), live satellite data versus ~6 h NWP lag, 100 m wind for turbines, and up to ~50% better precipitation; it powers Search, Gemini, Maps, and Earth Engine.
On September 8, 2026, Google DeepMind unveiled AlphaGenome Atlas with predictions for 9 billion single-nucleotide variants—every possible single-letter change in the human genome. The AVI score combines AlphaGenome and AlphaMissense; an academic portal, API, and Google Antigravity skill ship now; non-commercial web today, commercial Cloud soon.