Demis Hassabis, CEO, and co-founder of DeepMind, a London-based machine learning AI startup established in 2010, partnered with Shane Legg and Mustafa Suleyman to embark on a groundbreaking journey. Hassabis crossed paths with Legg during their postdoc days at the Gatsby Computational Neuroscience Unit, while his friendship with Suleyman, an Elixir partner, dates back to their family connections. Along the way, Hassabis also enlisted his university friend and Elixir partner, David Silver.
DeepMind’s overarching mission is to “solve intelligence” and employ that intelligence to address a wide array of challenges. To concretize this vision, DeepMind seeks to merge insights from systems neuroscience with cutting-edge developments in machine learning and hardware. This approach aims to unlock increasingly potent, versatile learning algorithms, eventually leading to the creation of artificial general intelligence (AGI).
DeepMind’s early investors included prominent tech entrepreneurs, and in 2014, Google acquired the company for £400 million. Although DeepMind largely retained its independence, DeepMind Health became integrated into Google Health.
Following the Google acquisition, DeepMind achieved numerous milestones, including the creation of AlphaGo, a program that defeated world champion Lee Sedol in the intricate game of Go, previously deemed a formidable AI challenge. Additional achievements encompass innovations such as the Neural Turing Machine, a 40% reduction in energy consumption for Google data center cooling systems, advancements in AI safety research, and a collaboration with the UK’s National Health Service (NHS) and Moorfields Eye Hospital to enhance healthcare and detect degenerative eye conditions.
More recently, DeepMind directed its AI prowess toward protein folding, a long-standing scientific challenge. In December 2018, DeepMind’s AlphaFold secured victory in the Critical Assessment of Techniques for Protein Structure Prediction (CASP), accurately predicting the structures of 25 out of 43 proteins, signifying a substantial step forward in this vital biological endeavor.
DeepMind’s contributions extend beyond these achievements, with significant technical advancements in machine learning, especially in deep learning and reinforcement learning. DeepMind also pioneered deep reinforcement learning, a fusion of these two techniques, driving innovations in the field.
Hassabis envisions AI as a profoundly beneficial technology for humanity but acknowledges the ethical challenges that accompany its advancement. He is among those who signed a statement emphasizing the global significance of addressing the risks associated with AI and the need for evaluation tests to assess the capability and controllability of new AI models. Hassabis believes that AI can play a vital role in addressing critical issues such as health and climate change, making it imperative to continue its development.
