Google DeepMind launches an AGI institute and opens with an economics essay graded by 51 AI raters
TechCrunch covered the DeepMind Institute launch on 17 September, a venue for Google and Google DeepMind researchers to publish AGI ideas explicitly flagged as not Google's official view. Its opening essay, 'Economic policy for AGI' by Julian Jacobs and Alex Imas, evaluates 11 policies across welfare, agency, feasibility and durability, and tiers them by severity: expanded unemployment insurance and an enhanced EITC for mild disruption, an EITC converted to a Negative Income Tax for moderate displacement, and Universal Basic Capital granting direct ownership stakes in publicly managed portfolios as the structural backstop. The methodology is the part worth arguing about, the authors used 51 AI agent raters trained on real economists to score the policies, alongside survey data showing 85% American support for retraining and 72% for expanded unemployment insurance.
↳ Follow the thread