We scan new podcasts and send you the top 5 insights daily.
Socher says economics is still ideological and built on one-step models with beautiful math but no adaptation. His AI Economist used two-level reinforcement learning, with tax-and-subsidy proposals tested against self-interested agents, to simulate policy outcomes. He wants policy recommendations made more objectively in the future.
The math used for training AI—minimizing the gap between an internal model and external reality—also governs economics. Successful economic agents (individuals, companies, societies) are those with the most accurate internal maps of reality, allowing them to better predict outcomes and persist over time.
Emad Mostaque proposes that the math behind generative AI can describe economic systems. In this framework, Adam Smith's theories map to "gradient flows" (scarcity), Marx's to "circular flows" (compounding intelligence), and Hayek's to "harmonic flows" (structural rules).
AI's greatest impact on economics will be the ability to run complex, agent-based simulations. This allows economists to model the dynamic, equilibrium responses of millions of economic actors to policy changes—like a Fed balance sheet reduction—providing a much richer understanding than traditional, static models allow.
Demis Hassabis foresees AI enabling new scientific disciplines. He suggests that highly accurate AI simulations could transform fields like economics into hard sciences by allowing for the kind of repeated, controlled experiments that are currently impossible in the real world.
AI excels at solving problems with clear, verifiable answers, like advanced math, allowing for effective training. It struggles with complex societal issues like unemployment because there is no single, universally agreed-upon "correct" solution to train against, making it difficult to evaluate the AI's path.
As AI automates the technical and computational 'science' of economics, the value of human economists will shift to the 'social' side. Future contributions will rely on the art of identifying important questions, exercising judgment in model design, and providing novel, real-world insights that AI cannot replicate on its own.
Socher describes his 2018 'AI Economist' project, which used multi-agent simulations to model an economy. This approach allows for testing different fiscal policies (like taxation) over billions of simulated scenarios to find the optimal strategy for a stated goal, removing partisan bias from policy-making.
Citing historical failures like David Ricardo's on automation, individual AGI forecasts are deemed useless. A better approach is to model potential scenarios (e.g., labor share collapses) and then identify the crucial, currently missing data (like consumer demand elasticities) needed to determine which scenario is likely.
Beyond automating tasks, Emad Mostaque's "Intelligence Theory" suggests AI's deepest impact is shifting the foundational axiom of economics. Instead of scarcity, the new core principle is persistence: how complex systems (like firms or AIs) maintain themselves by accurately modeling and predicting reality.
For a period, a perverse norm developed in economics where the 'better' academic model was one whose theoretical agents were smarter and more rational. This created a competition to move further away from actual human behavior, valuing mathematical elegance and theoretical intelligence over practical, real-world applicability.