Google DeepMind

News Summary:

Google DeepMind attracted academic talent, with at least 22 professors and researchers leaving or taking leave from elite universities including Stanford, Berkeley, and Harvard during roughly the first half of 2026 to join Google DeepMind and other leading AI companies, consolidating scarce academic expertise. Earlier on July 9, Google Cloud launched AlphaEvolve, an AI system developed with Google DeepMind, making it broadly available to Cloud customers for enterprise algorithm-optimization work in areas such as chip design, logistics networks, and medical research. This initiative aims to transform the search for improved algorithms into a consumable cloud capability. Concurrently, the AI industry is shifting, with Google DeepMind and competitors moving beyond traditional general-purpose models to focus on task-specific AI models like Sol, Sonnet 5, and Gemini 3.1, aiming for faster, more accurate, and cost-efficient performance. Earlier that day, Demis Hassabis continued to navigate corporate constraints at Google DeepMind, with its Gemini AI assistant having not reached the public awareness level of rivals, a dynamic attributed to the innovator's dilemma within large companies. Also on July 9, Google DeepMind introduced the Predicting the Past Skill for Google Antigravity, an AI research tool built with Thea Sommerschield from Durham University, designed to help historians and epigraphers analyze damaged ancient inscriptions through natural language by connecting Gemini with its specialist models Ithaca and Aeneas.