Google Deepmind has established the Deepmind Institute (DMI) to coordinate research and debate concerning artificial general intelligence. This new platform unites staff from Google Deepmind and Google with the wider scientific community to examine safety, governance, and risks such as cyberattacks or loss of control. Directors Shane Legg, James Manyika, and Nobel laureate Demis Hassabis state that answers must come from technologists alongside experts in the arts, humanities, and policy. The institute aims to address questions that cannot be solved by engineers working in isolation.
The launch highlights a lack of consensus on what AGI actually is. Deepmind defines the term as a system possessing all the cognitive abilities of the human brain and suggests it is near, despite current systems failing at simple tasks. Hassabis predicts arrival within a few years while Legg recently suggested a precursor could appear by 2028. OpenAI CEO Sam Altman believes it could happen by the end of this year, focusing on economic value rather than strict definitions. The group distinguishes AGI from artificial superintelligence, noting the latter would surpass humans in every respect but remains a separate stage. The initiative seeks to ground these predictions in broader societal impact rather than purely technical metrics.
* DMI brings together researchers from Google Deepmind, Google, and the global scientific community.
* Directors include Shane Legg, James Manyika, and Nobel laureate Demis Hassabis.
* Focus areas include safety, governance, and risks like cyberattacks or loss of control.
