Google co-founder Sergey Brin presses AI team to improve Gemini as Deep Mind reshuffles leadership
Google co-founder Sergey Brin has recently personally pressured the company’s AI team to devote more effort to improving the capabilities of its flagship Gemini models. People familiar with the matter said Google’s senior management is taking a more direct approach to overseeing DeepMind, pushing engineers to release new models faster and placing greater emphasis on commercial deployment. The leadership reshuffle highlights a shift toward commercialization. Last week, Google CEO Sundar Pichai announced major changes to the company’s AI leadership. DeepMind CEO Demis Hassabis will step down as CEO and become the lab’s chairman, while also taking on the role of chief scientist at Alphabet. He will continue to lead Isomorphic, DeepMind’s AI drug-discovery lab. Meanwhile, DeepMind’s current chief technology officer, Koray Kavukcuoglu, will be promoted to senior vice president of DeepMind and report directly to Pichai. The changes are closely tied to Google’s strategy of commercializing AI. Brin has also specifically urged the team to focus more on “recursive self-improvement”—the idea of having AI systems help design, write code for or train the next generation of models, thereby producing continuing, compounding gains in capability. AI-assisted coding has become one of the most foundational and revenue-generating areas of application. Last week, Google was reportedly in talks with San Francisco startup Mechanize, which focuses on building environments and evaluation systems for advanced coding agents. Brin and Larry Page co-founded Google in 1998. Although they stepped away from day-to-day management in 2019, they remain board members and continue to exert significant influence over the company as controlling shareholders. From Brin’s direct push for faster progress to DeepMind’s leadership reporting directly to the CEO, Google is clearly accelerating the pace of AI research and commercialization. Improving Gemini’s capabilities has been elevated to a higher priority.
The leadership reshuffle highlights a shift toward commercialization. Last week, Google CEO Sundar Pichai announced major changes to the company’s AI leadership. DeepMind CEO Demis Hassabis will step down as CEO and become the lab’s chairman, while also taking on the role of chief scientist at Alphabet. He will continue to lead Isomorphic, DeepMind’s AI drug-discovery lab.
Meanwhile, DeepMind’s current chief technology officer, Koray Kavukcuoglu, will be promoted to senior vice president of DeepMind and report directly to Pichai.
The changes are closely tied to Google’s strategy of commercializing AI. Brin has also specifically urged the team to focus more on “recursive self-improvement”—the idea of having AI systems help design, write code for or train the next generation of models, thereby producing continuing, compounding gains in capability.
AI-assisted coding has become one of the most foundational and revenue-generating areas of application. Last week, Google was reportedly in talks with San Francisco startup Mechanize, which focuses on building environments and evaluation systems for advanced coding agents.
Brin and Larry Page co-founded Google in 1998. Although they stepped away from day-to-day management in 2019, they remain board members and continue to exert significant influence over the company as controlling shareholders.
From Brin’s direct push for faster progress to DeepMind’s leadership reporting directly to the CEO, Google is clearly accelerating the pace of AI research and commercialization. Improving Gemini’s capabilities has been elevated to a higher priority.