The landscape of scientific discovery is shifting fast, as two technology giants make significant moves in AI-driven research and data collection. Google has announced the launch of Gemini for Science, a collection of experimental AI tools in Google Labs, while Uber is ramping up its autonomous vehicle ambitions with its new AV Lab project, aiming to generate 2 million miles of driving data each month by the end of 2026.
Google’s Gemini for Science: AI as a Scientific “Force Multiplier”
Starting this month, Google is rolling out access to Gemini for Science, a toolbox designed to accelerate and amplify researchers’ efforts across disciplines. The suite includes three main experimental tools: Literature Insights, Hypothesis Generation, and Computational Discovery. Built on top of Google DeepMind technologies like Co-Scientist, AlphaEvolve, and Empirical Research Assistance (ERA), these tools are intended to streamline everything from literature review and hypothesis development to computational modeling and code testing.
Literature Insights, powered by NotebookLM, helps scientists sift through mountains of academic papers, organizing them into searchable tables and enabling conversational analysis. Hypothesis Generation uses a multi-agent “idea tournament” to generate, debate, and evaluate new research questions, while Computational Discovery tests thousands of code variations in parallel—already being applied in fields like solar forecasting and epidemiology.
Google says more than 100 institutions, including Stanford University and Imperial College London, are already involved in validating these tools. Early tests have even led to potential breakthroughs in rare genetic disease research related to AK2 gene mutations. The company is also partnering with major scientific conferences to explore AI-assisted peer review and validation.
Uber’s AV Lab: Data-Driven Ambitions
Meanwhile, Uber is putting its autonomous vehicles back on the road—not as self-driving cars yet, but as data-collection workhorses. Each vehicle, equipped with cameras, lidar, and radar, will be manually driven for now but will generate valuable data for Uber’s dozens of robotaxi partners. Balaji Krishnamurthy, Uber’s CFO, stated that the fleet will reach a target of at least 2 million miles per month by the end of 2026, with plans to scale even further in 2027. This is a crucial step, as experts estimate AV operators need upwards of 10 million miles of driving data before public driverless launches can be considered safe.
The return to the AV field comes after Uber sold its original self-driving division in 2020, following a tragic fatality in Tempe, Arizona. With its new AV Lab, Uber is taking a more measured approach, focusing on data generation and safety as it rebuilds its ambitions in the autonomous vehicle space.
As AI accelerates both biological and technological advances, from modeling proteins to driving cars, one thing is clear: the intersection of data, machine learning, and human ingenuity is ushering in a new era for science and industry alike.
