DeepMind establishes new institute for cross-disciplinary AGI research

Google DeepMind has founded the DeepMind Institute (DMI) as a platform for interdisciplinary research on artificial general intelligence. The institute will bring together technologists with experts from the arts, humanities, and policy to examine safety, governance, and control risks. DeepMind leaders say AGI is near, though they acknowledge current systems still lack creativity and fail at simple tasks.

The DeepMind Institute will operate as a collaborative hub, pairing DeepMind’s technical researchers with outside scholars from disciplines such as philosophy, sociology, and public policy. Its stated mission is to broaden the conversation on AGI beyond engineering, addressing questions of accountability, cyber resilience, and the potential for unintended behavior. The institute’s leadership includes prominent figures like Demis Hassabis and Shane Legg, who have publicly projected AGI’s arrival within a few years, though they concede current models remain brittle and unoriginal. Notably, the institute’s founding follows a period of intense debate over AGI definitions, with rival organizations like OpenAI emphasizing economic metrics rather than human-like cognition.