How Deep Learning Can Give Birth to General Artificial Intelligence

From chessbots to chatbots and from search algorithms to industrial assembly machines – humanity has arrived at the age of machine intelligence; or at least machine work, if we take seriously a definition of intelligence as “the abilities to learn from experience, adapt to new situations, understand and handle abstract concepts, and use knowledge to manipulate one’s environment“ (Encyclopedia Britannica). While some machines like IBM’s Watson are astonishingly good at solving certain problems, a general artificial intelligence (AI) which would satisfy the definition above has yet to be created. AI research has been around since the mid-fifties, but the enormous scientific and technical challenges of building a “full AI” are still keeping researchers and programmers awake at night. Spearheading today’s research is a London-based start-up, DeepMind, which was founded in 2010 and bought by Google in 2014. Co-founder and CEO Demis Hassabis is a former child chess prodigy, video game creator, neuroscientist, and an artificial intelligence researcher whose vision is to “solve intelligence and use it to solve everything else”. With more than 150 brilliant machine learning experts, AI researchers and coders, Google DeepMind is now one of the most promising candidates for realizing this extremely ambitious goal. At Falling Walls, Demis provides rare insights into the organisation that is set out to create an “Apollo mission” for AI research.

Demis Hassabis

Google DeepMind

Artificial Intelligence (AI) has always been a Demis’ focus. In his early career, his efforts were mainly towards AI programming for video games (e.g. Theme Park, Black & White, and Republic: The Revolution). After that, he entered the world of cognitive neuroscience where he worked on autobiographical memory and amnesia.. His work on cognition allowed Demis to find inspiration from the brain for new algorithmic ideas for AI. In 2011, this led him to co-found DeepMind Technologies, a London-based machine learning startup. 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. On January 26th, Google acquired DeepMind Technologies for £400 million. Demis is currently Vice President of Engineering at Google DeepMind where he hopes to solve intelligence and then use it to help society solve everything else.