DeepMind launches institute to prepare society for AGI

The DeepMind Institute brings together researchers from Google and beyond to examine how society could adapt to artificial general intelligence. Led by Shane Legg, James Manyika and Demis Hassabis, it opens with proposals for responding to job disruption and preserving visibility into AI reasoning. An economics essay recommends scaling support to observed disruption; a safety essay warns that readable reasoning could disappear without deliberate protection. The directors say current AI still falls short of full AGI, and the institute stresses that its essays represent their authors’ ideas, not official Google policy.