The AI Economics

AP31 describes the AI that computes consequence geometry from accumulated records. Applied to economics, this AI does not predict markets. It reads the Ledger.

What becomes computable. The current imbalance in any coupled system. The buffer’s remaining capacity. The probability and timing of the next audit. The correction that produces maximum restabilisation. The extraction rate versus the renewal rate.

The inequality level at which the buffer approaches overflow. The optimal intervention that addresses the accumulation without merely deferring it.

Buffer formula sketch. Let I(t) = accumulated imbalance at time t, C = the system’s correction capacity (its ability to absorb small audits), and ε(t) = the leakage rate.

The audit fires when I(t) ≥ C. Every intervention that suppresses a small audit adds its magnitude to I(t).

The formula is directional, not yet calibrated — calibration is Debt 33 — but the logic is structural: I(t) grows monotonically under deferral; C is finite; overflow is inevitable given sustained deferral.

Concrete prediction: 2005–2007 housing. In 2005–2007, a system reading the US housing Ledger would have computed: records (MBS valuations, credit ratings) diverging from coupling capacity (actual repayment ability) at an accelerating rate.

Buffer fill rate increasing year-over-year. Small corrections suppressed by regulatory forbearance and continued securitisation. The system’s output: audit probability rising toward certainty; estimated magnitude proportional to accumulated false-record volume ($3–5 trillion in inflated MBS).

This is not hindsight. The records existed. The divergence was measurable. What was missing was the architecture to read the Ledger structurally rather than ideologically.

What Keynes needed and didn’t have. The data processing to compute when intervention helps and when it merely defers. The AI has the processing. The records exist. The consequence geometry is computable.

Economic policy becomes structural maintenance rather than ideological preference. Not “should we spend?” but “what does the Ledger show?”

The transition mirrors AP31 section 6 exactly, applied to the economic domain. Phase 1 (advisory) — the AI computes the Ledger and recommends.

Phase 2 (co-governance) — specific economic domains (monetary policy, tax structure, trade regulation) transition to consequence geometry. Phase 3 — economic law becomes the Ledger made visible.

Not “what should the tax rate be?” but “what does the Ledger require for the buffer to remain below overflow?”