When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
What is Google DeepMind?
Google DeepMind is the research unit within Google AI that helped develop and release Gemini, Google’s first multimodal generative AI chatbot. Google CEO Sundar Pichai has said that the Gemini large language model (LLM) family fulfills the vision Google had when they formed Google DeepMind.
According to Sundar Pichai:
“These are the first models of the Gemini era and the first realization of the vision we had when we formed Google DeepMind earlier this year.”
Techopedia Explains Google Deepmind
In 2014, Google acquired DeepMind Technologies Limited, an independent artificial intelligence (AI) research lab in London. The purchase was considered to be a strong investment because, among other things, the AI lab was making progress using game theory and reinforcement learning to create a program that could potentially win a highly complex board game called Go.
After the acquisition, DeepMind Technologies Limited continued to operate under its original name, and the AI lab retained a level of independence until 2023. In April 2023, Google announced that Google Brain and DeepMind Technologies would merge, and researchers and developers at both labs would become a single team under the direction of DeepMind Technologies’ CEO, Demis Hassabis.
From now on, the new unit will be referred to as Google DeepMind.
AI Milestones
Since its inception, DeepMind has continually pushed the boundaries of what artificial intelligence can do. Here are some of the research lab’s most notable achievements:
2016
DeepMind Technologies captured wide public interest when AlphaGo defeated Lee Sedol in a five-game match. This was a landmark moment in neural networks and machine learning (ML), because it showcased how deep learning systems could use reinforcement algorithms and supervised learning techniques to learn how to perform complex cognitive tasks.
2017
A second version of the game called AlphaGo Zero beat the original AlphaGo in a series of 100 games. AlphaGo Zero also used reinforcement learning, but this time the AI programming was not given training data. Essentially, AlphaGo Zero played Go against itself and used game outcomes to build a knowledge base that helped it learn how to win the game.
This achievement is considered a major milestone in artificial intelligence because it demonstrated the potential for AI systems to learn, adapt, and solve complex problems autonomously without requiring training data.
2018 – 2020
DeepMind Technologies created AlphaFold in 2018 and unveiled AlphaFold 2 in late 2020. Both systems were designed to predict the 3D structure of proteins, which is essential for understanding biological processes and disease mechanisms.
AlphaFold 2’s ability to accurately predict protein structures within days had a significant impact on biochemistry and molecular biology research. (Before AlphaFold, it took months or even years to determine a protein’s structure.)
2021
DeepMind Technologies released the source code for AlphaFold2. It also released a database of protein structure predictions generated by AlphaFold and made it freely available to the scientific community.
April 2023
Google consolidated Google Brain and DeepMind Technologies resources and expertise and formally launched Google DeepMind. Google AI’s new unit, which is referred to internally as GDM, is expected to accelerate AI research while supporting the principles of responsible AI.
September 2023
Google DeepMind released AlphaMissense AI, an AI tool that analyzes genetic mutations and predicts their potential impact on health. AlphaMissense AI has the potential to significantly contribute to faster and more accurate diagnostics for some genetic diseases and assist in the development of personalized healthcare treatments.
