Company Profile: Who Is DeepMind?
We take a closer look into the artificial intelligence giant, DeepMind, and how it has come to be so successful in its industry.
DeepMind Technologies is a UK artificial intelligence company founded in September 2010, and acquired by Google in 2014. The company is based in London, with research centres in Canada, France, and the United States. In 2015, it became a wholly-owned subsidiary of Alphabet Inc.
The company has created a neural network that learns how to play video games in a fashion similar to that of humans, as well as a Neural Turing machine, or a neural network that may be able to access an external memory like a conventional Turing machine, resulting in a computer that mimics the short-term memory of the human brain.
DeepMind Technologies' goal is to "solve intelligence", which they are trying to achieve by combining "the best techniques from machine learning and systems neuroscience to build powerful general-purpose learning algorithms". They are trying to formalize intelligence in order to not only implement it into machines, but also understand the human brain.
The company first came to public attention in 2016, when DeepMind’s AlphaGo won a best-of-five against Go world champion Lee Sedol. Though chess-playing supercomputers such as IBM’s Deep Blue have been around since the 90s, DeepMind approached the massively more complex game of Go with machine learning techniques including neural networks and reinforcement and imitation learning.
Its impact so far:
Over 100 million people are affected by diabetic retinopathy or age-related macular degeneration. These conditions can cause permanent sight loss unless they’re treated quickly. The results, which were published in Nature Medicine, showed that their AI system could recommend patient referrals as accurately as world-leading expert doctors for over 50 sight-threatening eye diseases. More recently, they showed that the intelligent system can predict whether a patient will develop a more severe form of age-related macular degeneration months before it happens–paving the way for future research in sight-loss prevention.
Knowing how proteins fold to create different shapes could help scientists understand a protein’s role within the body. This discovery might help treat diseases believed to involve misfolded proteins such as Parkinson’s, Huntington’s and cystic fibrosis. Predicting the shape of proteins is a major unsolved challenge in science and the company has already seen early signs that their AI systems could accelerate progress in this field.
DeepMind's teams working on technical safety, ethics, and public engagement aim to address these questions and more. They help to anticipate short and long-term risks, explore ways to prevent these risks from happening, and find ways to address them if they do.
They believe this approach also means ruling out the use of AI technology in certain fields. For example, they have signed public pledges against using their technologies for lethal autonomous weapons, alongside many others from the AI community.
These issues go well beyond any one organisation. DeepMind's ethics team works with many brilliant non-profits, academics, and other companies, and creates forums for the public to explore some of the toughest issues. The safety team also collaborates with other leading research labs, including our colleagues at Google, OpenAI, the Alan Turing Institute, and elsewhere.
It’s also important that the people building AI reflect the broader society. They are working with universities on scholarships for people from underrepresented backgrounds, and support community efforts such as Women in Machine Learning and the African Deep Learning Indaba.
DeepMind and coronavirus:
One of its latest projects involves turning its technology on to the study of coronavirus. The company’s AlphaFold system analyses protein structure and folding.
In a blog post, the company explained the application of its system to the virus. “AlphaFold, our recently published deep learning system, focuses on predicting protein structure accurately when no structures of similar proteins are available, called “free modelling”. We’ve continued to improve these methods since that publication and want to provide the most useful predictions, so we’re sharing predicted structures for some of the proteins in SARS-CoV-2 generated using our newly-developed methods.”
Find out more about the company, here.