Google technologies now support more than 300 languages, spoken by 7 billion people - 86% of the global population - a milestone the company announced alongside a raft of AI advances in science and crisis prediction. The updates, detailed in a September 15 blog post, show AI models already in use for disease screening, weather forecasting, and economic analysis across multiple countries.
The company also released the AI & Economy ATLAS, an interactive dataset showing how people globally use AI at work and in daily life. The announcements come with a clear focus: applying AI to accelerate scientific discovery in ways that directly improve health, resilience, and economic opportunity.
Health and genomics tools reach new scale
Google's AlphaGenome Atlas now maps all 9 billion possible single-letter genetic changes across the human genome, with the data made openly available to researchers. It builds on AlphaFold, which has predicted all 200 million protein structures known to science and is used by 4 million researchers in 190 countries. AlphaMissense, another tool in the suite, helps predict disease-causing genetic mutations.
In clinical settings, a breast cancer study with Imperial College London and the UK's NHS showed AI detected 25% of interval cancers previously missed in mammograms of 175,000 women. For tuberculosis, where roughly 40% of infected people worldwide go undiagnosed, a chest X-ray tool has screened over 25,000 images across 40 locations in six nations. A diabetic retinopathy model, developed with partners, has supported more than 1.15 million screenings globally, with plans to reach 6 million over the next decade.
On the research side, collaborative AI tools like Co-Scientist are helping scientists generate and validate novel hypotheses, including identifying new therapeutic applications for existing drugs for acute myeloid leukemia. Open-sourced tools such as DeepConsensus, DeepVariant, and DeepPolisher have assisted in completing the human genome and drafting the first pangenome through the Human Pangenome Reference Consortium.
Weather and crisis prediction models improve accuracy
WeatherNext 3, Google's newest global weather model, delivers 50% more accurate precipitation forecasts a day or more ahead. It combines real-time satellite observations with AI to produce high-resolution, hourly forecasts without immense supercomputing power - a step forward for data-sparse regions long underserved by high-resolution forecasting. Last year, authorities in Jamaica used WeatherNext to predict Hurricane Melissa's path, securing early disaster funding before the storm made landfall.
Flood Hub now covers 2 billion people across more than 150 countries, including both riverine floods and flash floods. In 2025, monsoon predictions provided information for 38 million farmers in India. Wildfire boundary predictions operate in the U.S. and 33 other countries, while a FireSat satellite constellation project aims to detect wildfires previously too small to spot. In 2025, Google generated more than 520 crisis alerts on Search, reaching over 75 million users with timely wildfire information.
The Earth AI Planetary Prediction Engine (PPE), an autonomous AI system, uses plain-language instructions to forecast crises like disease outbreaks, food shortages, and climate risks. During the ongoing Ebola outbreak in the Democratic Republic of the Congo, PPE pinpointed 83% of emerging hotspots ahead of time, outperforming current forecasting systems. In Nigeria, it doubled food security forecasting accuracy at the local district level, and in the U.S. it outperformed traditional models in identifying vulnerable communities across 21 CDC health indicators.
Language access and economic insights expand
Google Translate now supports nearly 300 languages, helping more than 1 billion people translate around 1 trillion words each month. The company's goal is to support the world's 1,000 most-spoken languages. Partnerships with Makerere University, University of Ghana, and Digital Umuganda produced an open dataset for 27 African languages spoken by over 100 million people. Project Vaani, with IISC-Bangalore and ARTPARK, open-sourced speech and image datasets for 109 Indic languages. A sign-language-to-text model (SL2T) now enables sign-to-text dictation.
On the economic front, the AI & Economy ATLAS provides empirical insights on how people use AI tools at work. Google has provided more than $1 billion globally in training and skilling initiatives, helping over 100 million people gain digital and AI skills. The company also gave more than $1 billion across 1,700-plus research institutions worldwide since 2006 to support academic science and discovery.
Why this matters for science and research professionals
The tools and datasets described here are not theoretical - they are already deployed and generating results in peer-reviewed studies, clinical settings, and crisis response operations. For research scientists, the open availability of resources like AlphaGenome Atlas and the pangenome toolchain means new starting points for hypothesis generation and validation. The AI Learning Path for Research Scientists addresses the skills needed to work with these kinds of models. As the blog post puts it, AI creates "a positive feedback loop that can keep growing with no end in sight" - and the infrastructure to participate in that loop is increasingly public and production-ready. For professionals in AI for Science & Research, the signal is clear: the gap between model release and real-world impact is shrinking fast.
Your membership also unlocks:
