Google DeepMind Establishes DeepMind Institute (DMI)

Google DeepMind launched the DeepMind Institute (DMI), a cross-disciplinary platform for AGI research. Co-founded by Shane Legg, Demis Hassabis, and James Manyika, DMI will publish deep thinking on AGI's benefits and risks.
Key insights
DMI Structure and Purpose
DMI is a cross-disciplinary publishing platform for deep thinking about an AGI world, distinct from Google's official stance and not a pure technical club. It is co-founded by Shane Legg (chief AGI scientist), Demis Hassabis (Nobel laureate, Alphabet chief scientist), and James Manyika (president of research and technology society), with Legg serving as editor-in-chief.
AGI Definition and Timeline
The institute defines AGI as a system exhibiting the full cognitive capabilities of the human brain. It states that current AI still fails at basic tasks and lacks consistency and true creativity, but these gaps are expected to be filled soon, indicating AGI is approaching.
Civilization-Level Benefits and Risks
AGI is described as the ultimate tool for accelerating scientific breakthroughs, including curing diseases, clean energy, and universal prosperity. However, risks are equally serious, specifically naming cybersecurity, biological risks, and the potential for uncontrolled self-improving systems. The institute emphasizes the current window to ensure benefits outweigh risks.
Founders' Long-Term Vision
Demis Hassabis and Shane Legg have discussed AGI's potential impact on economy, science, and society for over 20 years. DMI expands interdisciplinary research on key questions for the AI era, aiming to spur discussions needed to get the next steps right.