Google DeepMind Launches Institute to Explore Artificial General Intelligence
New research platform will examine AGI’s potential impact on society, economy, safety and policy
Technology Desk: Artificial intelligence has rapidly become part of everyday life, but the concept of artificial general intelligence, or AGI, remains less familiar to the wider public. Seeking to expand research and public discussion around the emerging technology, Google and Google DeepMind have launched the DeepMind Institute.
The institute was launched on September 16 as a research and publishing platform focused on the potential opportunities and challenges associated with Artificial General Intelligence (AGI). Its contributors will include AI researchers as well as economists, policymakers, social scientists and thinkers from other fields.
Rather than developing a new AI model or consumer-facing software, the institute will focus on research, analysis and debate surrounding the future of advanced AI.
What is AGI?
Artificial general intelligence refers to a hypothetical AI system capable of performing a broad range of cognitive tasks rather than being limited to a specific function.
Current AI systems can perform tasks such as generating images, understanding language, writing code and analyzing information. AGI, by contrast, is generally described as a system that could learn from new situations, reason across different domains and apply knowledge to unfamiliar problems in ways comparable to human intelligence.
Google DeepMind has said that although the capabilities of current AI systems are advancing rapidly, they have not yet reached the level generally associated with full Artificial General Intelligence (AGI). Continued technological development, however, could reduce some of the limitations of existing systems.
What will the DeepMind Institute do?
The DeepMind Institute is intended to provide a platform for research into how advanced AI and AGI could be developed and managed safely.
Its research will examine issues including the potential social and economic consequences of Artificial General Intelligence (AGI), AI safety, human well-being and the policies and regulations that may be needed as AI capabilities continue to advance.
The institute is led by DeepMind co-founder Shane Legg, Google executive James Manyika and Google DeepMind CEO Demis Hassabis. Legg is also serving as managing editor.
By bringing researchers from different disciplines together, the institute aims to broaden the AGI debate beyond the technology sector and examine its potential implications for society and the economy.
Four research papers published at launch
Four research papers were released when the institute launched.
One examines policy measures that could be needed to address potential economic changes associated with AGI. Another explores how observable the internal reasoning, often referred to as the “chain of thought,” of advanced AI models can be to humans.
The other publications examine principles for human well-being in an era of Artificial General Intelligence (AGI) and frameworks for evaluating the capabilities of advanced AI systems.
One of the papers also considers the potential safety and oversight challenges that could arise if researchers have limited visibility into the internal reasoning processes of increasingly advanced AI models.
Separately, Demis Hassabis has proposed establishing an independent standards-setting organization, led by the United States, to evaluate the capabilities and safety of advanced AI models before they are released to the market.
Research available to the public
The DeepMind Institute is already online, with its published research available to readers free of charge. Researchers, technology professionals, academics, economists, policymakers and members of the public interested in the future of AGI can access the material.
The institute also emphasizes that its publications should not be interpreted as official Google positions. Individual authors may revise their views as new evidence and research emerge.
With AGI still a developing concept, its potential impact is being discussed across technology, economics, public policy and social science. The DeepMind Institute is intended to provide a forum for bringing those different perspectives into a broader conversation about the opportunities, risks and governance of advanced artificial intelligence.
