Google DeepMind

Google's main AI research lab, founded in London in 2010. It builds the Gemini models and science systems such as AlphaFold, with the mission of building AI responsibly to benefit humanity.

About Google DeepMind

Google DeepMind is Google's main AI research lab and one of the best-known AI teams in the world. Its scientists, engineers and ethicists build AI systems and apply them to hard problems in science and engineering. It is behind Gemini, Google's family of AI models that can work with text, images, audio and video, and behind research projects that made headlines worldwide, from beating champions at the board game Go to predicting the shapes of proteins.

The story began in London in 2010, when Demis Hassabis, Shane Legg and Mustafa Suleyman founded DeepMind. Hassabis and Legg had met at the Gatsby Computational Neuroscience Unit at University College London. Their early work taught AI to play classic Atari games such as Breakout and Pong, and their aim was to build a general-purpose AI that could be useful for almost anything. Google bought the company in 2014, and in 2023 DeepMind joined forces with Google Brain, Google's other AI research team, to become Google DeepMind. It is headquartered at King's Cross in London and, according to Wikipedia, also has research centres in the US, Canada, France, Germany and Switzerland.

Google DeepMind states its mission as building AI responsibly to benefit humanity. Its goals include advancing science, changing how people work, serving diverse communities and improving billions of people's lives, with the long-term ambition of developing advanced AI, often called artificial general intelligence, safely. It describes a holistic approach to responsibility and safety, and combines researchers with ethicists and policy experts.

Several of its projects are famous. AlphaGo beat the world-class Go player Lee Sedol in 2016. AlphaFold predicts the 3D structure of proteins, work that led Demis Hassabis and his colleague John Jumper to share the 2024 Nobel Prize in Chemistry. Today the lab builds Gemini; Veo, which creates video; Genie, which generates interactive worlds; Gemini Robotics; Gemma, a set of open models anyone can use; and tools for science such as weather forecasting (WeatherNext) and genome analysis (AlphaGenome).

The lab says it works by the scientific method, with teams that mix scientists, engineers and ethicists on shared problems, and it highlights real-world impact for careers. The pages we read do not describe a specific internship or graduate scheme, so students should check the live roles on the Google DeepMind careers page.

What they do

• {'desc': "Google's family of AI models for text, images, audio, video and code.", 'title': 'Gemini models'}
• {'desc': 'AI for biology, weather, materials and mathematics, including AlphaFold.', 'title': 'Science and discovery'}
• {'desc': 'Models that make video (Veo), music (Lyria) and interactive worlds (Genie).', 'title': 'Generative media'}
• {'desc': 'Gemini Robotics and work on AI that can act in the physical world.', 'title': 'Robotics'}
• {'desc': 'Gemma, a family of openly available AI models for developers.', 'title': 'Open models'}
• {'desc': 'Research and policy work on making advanced AI safe and beneficial.', 'title': 'Responsibility and safety'}

Key facts

• Founded: London, 2010
• Headquarters: King's Cross, London, UK
• Founders: Demis Hassabis, Shane Legg, Mustafa Suleyman
• Part of Alphabet (Google) since 2014
• Hassabis and John Jumper shared the 2024 Nobel Prize in Chemistry for AlphaFold

Website: deepmind.google