AInomicsThe Economics of Choice in the Age of AI—Governing Superintelligence, Taxing AI Capital, and Sharing the Value of Data Public
HUMAN · AI · POLICYMAKER

Choices connect.
Futures change.

What society values. What the world brings.
Explore how social priorities and unexpected shocks shape policy and future paths.

AInomics: The Universal Math of Choice. Generative AI, physical AI, humans, central banks, and governments optimize choices under uncertainty. Stochastic control, CGE and SDM, ABM, behavior, and political preferences connect people, the economy, and policy toward a prosperous and sustainable future. In the AI era, we must reimagine economics of choice.
THE UNIVERSAL MATH OF CHOICE
40 quarters · 3 states · 2 policy tools
Economy Flight Simulator — Stochastic Optimal Control in a DSGE Model with Sector-Specific Political Preferences. A human pilot and an AI copilot steer an aircraft above a city, with cockpit displays for central bank monetary policy and government fiscal policy.
Economy Flight Simulator · Stochastic Optimal ControlView full-size image
iIllustrative model · These are not estimates of the Korean economy. All state and policy results are standardized indices, not actual interest rates or fiscal amounts.
COMPARE THE PATHS

Same shocks, different choices

Baseline: prices 45%, output 30%, equity 25% · same shocks & noise
Common loss improvement
—
Against baseline · positive means improvement
Peak state deviation index
—
Maximum state vector magnitude
Frozen-weight stability
—
max ρ(A − BK) · below 1
SYSTEM RESPONSE

Economic & social state xt

Your scenarioBaseline
Inflation deviation from target · standardized index—

The vertical line marks the selected quarter. Shocks enter in Q5 and first affect the state in Q6.

POLICY RESPONSE

Policy paths ut

+ tightening / − easing—
HUMAN → AI

Reading preferences

Estimate ẑLatent zObserved y

Band: estimate of z ±1σ · dots: annual observations used by the filter

Selected quarterQ9 Year 3
Q40
↳

Loading scenario…

THE CLOSED LOOP

Five variables. One connected choice.

Select a node to inspect its values for the chosen quarter and its connection to the next step.

Q9 · Your scenario
HUMAN ↔ AI ↔ POLICYMAKER
SYSTEM STATEQ9

xₜ · Economic & social state

xₜ₊₁ = Axₜ + Buₜ + Cεₜ

AInomics: Next-Gen Behavioral Policy Architecture. Five connected stages: ECAPS captures beliefs, affect, preferences, and social context; ABM simulates heterogeneous agent interactions; social choice maps measured social preferences into policy weights; stochastic control computes optimal interventions; outcomes and AI interactions feed back into human behavior states.
Conceptual framework · Human behavior, social choice, and adaptive policyView full-size image
AInomics ECAPS Bayesian Parameter Recovery Test using PMCMC and a particle filter. Three panels compare posterior distributions, true parameter values, and estimated means for diagnostic overreaction bias, network peer influence, and observation noise. A fourth panel compares the hidden ECAPS state, noisy survey observations, and a normalized macro shock over time.
ECAPS Bayesian Parameter Recovery · PMCMC / Particle FilterView full-size image
Four panels show macroeconomic gaps and public fear during a stagflation shock; electoral cycles and winning party alignment; endogenous shifts in inflation, growth, inequality, and responsiveness weights; and vote shares for center-left, center-right, and populist parties across election cycles.
Macroeconomic Shocks, Electoral Cycles, and Policy WeightsView full-size image
AInomics Complete ABM–ECAPS–Kalman–LQR Closed-Loop Dynamics. Four panels show macroeconomic inflation, output, and inequality gaps; optimal monetary and fiscal control signals; raw and Kalman-filtered MSPI preference weights; and micro-ECAPS fear, cognitive capacity, and preference polarization over time, with the exogenous shock period highlighted.
ABM–ECAPS–Kalman–LQR Closed-Loop DynamicsView full-size image
THREE TIMESCALES

Respond quickly. Agree carefully.

Flexible policy. Careful estimation. A stable constitution.

↗EVERY QUARTER

Fast policy

Policy decisions · every quarter

Compute monetary and fiscal policy uₜ in response to the current state xₜ. Objective weights wₜ remain fixed until the next annual update.

uₜ = −K(wₜ)xₜ
◎EVERY YEAR

Medium estimation

Preference estimation · every 4 quarters

Predict each quarter and assimilate observations yₜ in Q1, Q5, Q9… Map the annual estimate ẑₜ through the constitutional rule to update wₜ.

yₜ → ẑₜ → wₜ = f(ẑₜ ; θ)
◇EVERY 4 YEARS

Slow constitution

Constitutional agreement · every 16 quarters

Hold constitutional parameters θ—base priorities and feedback sensitivity—fixed. Reviews do not trigger automatic revisions; a Q17 change is an experiment chosen in advance by the user.

Fixed θ ≠ fixed wₜ

Stability takes more than one number.

ρ(A − BK) < 1 diagnoses state control with fixed weights. Stability of the full system, including changing weights and preference feedback, requires further analysis.

MODEL NOTES