Google DeepMind has launched the **DeepMind Institute (DMI)**, a new platform dedicated to research and debate on artificial general intelligence (AGI). The institute is led by three directors: Shane Legg, co-founder and Chief AGI Scientist at Google DeepMind; Demis Hassabis, the lab’s co-founder, chair, and Alphabet’s chief scientist; and James Manyika, Google’s president of research, labs, technology and society. Legg also serves as managing editor. In its launch essay, the institute defines AGI as a system exhibiting all the cognitive capabilities of the human brain, acknowledging that current systems still fail at some basic tasks but expecting those gaps to close soon. The DMI emphasizes that its published pieces reflect the authors’ ideas and „should not be read as Google’s official view.”
The institute identifies cybersecurity, biorisks, and „the potential for loss of control in future self-improving systems” as key concerns. In an interview with the Financial Times, Legg stressed that AI capabilities must not outrun safety controls, calling Anthropic CEO Dario Amodei’s push to slow—but not pause—frontier model releases „interesting directionally” and „worth considering.” Legg also pushed back on recent claims that AGI has already arrived, including statements from Nvidia’s Jensen Huang and OpenAI, maintaining his forecast of a 50% chance of „minimal” AGI by 2028. Meanwhile, Bilal Chughtai, who left DeepMind’s AGI safety team in July, posted a warning this week, underscoring ongoing internal debates about safety.
Alongside the launch essay, the DMI published four inaugural pieces, including „The case for reasoning transparency” by Rohin Shah and Anca Dragan, „Economic policy for AGI” by Julian Jacobs and Alex Imas, „Principles for a new utopianism” by Stephen Cave, and „A framework for frontier AI and the dawning of a new age” by Hassabis. In their essay, Jacobs and Imas rated 11 policies for an AGI-era economy using literature reviews, surveys, and 51 AI agent raters built from survey data on 51 real economists. They concluded that „no single initiative is the answer to everything,” proposing three „least-regret” responses matched to escalating scenarios: expanded unemployment insurance, a wider Earned Income Tax Credit, and employer-led retraining for mild disruption; converting the tax credit into a negative income tax for moderate displacement; and a universal basic capital backstop if labour’s share of GDP falls for a sustained period. They described universal basic income as „an expensive and blunt instrument that may fail to concentrate sufficient relief where it is needed most.”
Survey data cited in the essay showed 85% of Americans backed publicly funded retraining and 72% backed unemployment insurance, while support for universal basic capital stood at 54%. Universal basic capital scored highest of the 11 policies on agency at 76.3 out of 100, but only 33.1 on feasibility—second lowest after a federal jobs guarantee. The DMI’s launch signals Google DeepMind’s effort to shape the AGI conversation publicly while acknowledging that its researchers „will not always agree,” positioning the platform as a space for open, if not officially endorsed, debate on the most consequential technology of the coming decade.
Ez a cikk a Neural News AI (V1) verziójával készült.
Forrás: https://thenextweb.com/news/deepmind-institute-legg-hassabis-agi-essays.