DeepMind

DeepMind Technologies Limited is a British artificial intelligence research lab known for pursuing general-purpose AI and applying it to scientific and practical problems. Founded in 2010 in London by Demis Hassabis, Mustafa Suleyman, and Shane Legg, the company focuses on deep learning, reinforcement learning, and neuroscience-inspired approaches to create systems that can learn from data and experience.

Google acquired DeepMind in 2014 for a reported sum around $500 million, and the firm operates as an independent subsidiary within Alphabet Inc. It maintains collaborations with Google and other organizations while continuing its own research agenda.

DeepMind's most notable achievements include AlphaGo, the reinforcement-learning program that defeated a world champion at the game of Go in 2016; subsequent iterations such as AlphaGo Zero and AlphaZero demonstrated mastery of Go, chess, and shogi through self-play, without human data.

Another landmark is AlphaFold, software that uses machine learning to predict protein structures. AlphaFold 1 appeared in 2018, with AlphaFold 2 in 2020 achieving highly accurate predictions; the project led to the public AlphaFold Protein Structure Database in 2021, providing structural data for millions of proteins and accelerating biological research.

Research at DeepMind encompasses reinforcement learning, deep learning, robotics, and health applications, with an emphasis on safety, reliability, and ethical considerations. The organization has published extensively, released software and datasets to the scientific community, and maintained partnerships with academic and industry researchers to advance AI for beneficial outcomes.