Google merges Brain and DeepMind to form Google DeepMind

Google has brought its two major artificial intelligence labs, Google Brain and DeepMind, under a new division called Google DeepMind. Demis Hassabis, DeepMind’s co-founder and CEO, will lead the group and report directly to Sundar Pichai, CEO of Alphabet and Google.

The move, announced Thursday, brings together researchers, engineers and computing infrastructure that had previously operated in separate organizations. Google is seeking to accelerate AI product development and scientific advances at a time when ChatGPT’s popularity and the launch of GPT-4 have intensified competitive pressure on the biggest technology companies.

Two AI stories under one roof

Google Brain began in 2011 as a small research project within Google. Over time, it became a central part of the company’s machine-learning strategy. Its work produced major contributions to TensorFlow, the open-source platform used to build and train AI models, as well as to translation, image-recognition and language systems.

DeepMind, founded in London in 2010 and acquired by Google in 2014, followed a different path. Its name became widely known through AlphaGo, the system that defeated Go champion Lee Sedol in 2016. More recently, AlphaFold predicted the three-dimensional structure of proteins with an accuracy that has had an impact on biology and drug design.

Both teams shared the ambition of building increasingly capable AI systems, but they had different cultures and priorities. Brain was more closely integrated with Google’s products, while DeepMind maintained a distinct identity for years, with a strong emphasis on fundamental research and AI safety.

The merger removes that separation. Google DeepMind will combine research, technical talent and computing capacity to work on both advanced models and applications for the company’s products.

Google’s response to the generative AI race

The decision comes just one month after the public launch of Bard, Google’s chatbot, initially based on LaMDA, its conversational language model. The company also introduced PaLM, a family of large language models, and has announced tests and plans to bring generative features to tools such as Gmail and Docs.

Generative AI produces text, images, code and other content from instructions written by a person. Its arrival has shifted the focus of competition: publishing research or improving invisible features within a product is no longer enough. Companies need to turn their models into reliable, scalable services that millions of users can understand.

Google has an uncommon advantage: data, cloud infrastructure, top-tier researchers and products with billions of users. But it also faces a challenge: bringing research into public-facing services requires controlling errors, harmful responses, computing costs and potential misuse. Bard launched on a limited basis in the United States and the United Kingdom, a cautious approach that contrasts with ChatGPT’s rapid expansion.

Demis Hassabis takes the helm

Hassabis will be the central figure in the new organization. A neuroscientist and video-game developer before founding DeepMind, he has argued for years that major scientific breakthroughs and artificial general intelligence—a machine capable of handling a wide range of tasks—require long-term research, not just immediate products.

The new unit will have to balance that vision with Google’s commercial needs. Jeff Dean, formerly head of Google Research and one of Brain’s historic leaders, will remain involved in the company’s research and technical ecosystem. The integration therefore does not mean abandoning Google’s scientific work; it means concentrating the core of its advanced AI efforts under a single leadership structure.

The results will be measured in upcoming launches. Google DeepMind will have to show that it can translate laboratory achievements into useful tools without sacrificing the caution demanded by a technology capable of influencing search, education, programming, science and office work. For Google, the merger is more than an organizational change: it is an attempt to turn decades of research into a faster response to AI’s new era.