6 months+0.93%
100.9282 99.8166101.7384
Deterministic shadow release · August 3, 2026
AI Productivity Index
AIPI turns AI progress into an expected productive-capacity term structure.
A single index, based at 100, estimates how interacting forces across compute, energy, infrastructure, financing, and adoption may translate into economy-wide productivity over time.
36-month AIPI106.8441 +6.84% from base
65% coverage57.75% confidence
Evidence quality is reported separately and never changes the index level.
Headline term structure
Expected gain, made legible across time
Central values are probability-weighted across six preserved scenarios.
12 months+2.55%
102.5472 99.5748104.9921
36 months+6.84%
106.8441 100.1346112.5721
Scientific status: illustrative and deterministic, not yet economically validated. This is not an investment benchmark, forecast product, or recommendation.
Scenario field · 36 months
Uncertainty stays visible
AIPI does not collapse the future into one unjustified narrative. Six named branches remain inspectable beneath the scalar, including bottlenecks, interruption, adoption, and demand expansion.
Baseline absorption30% prior
105.7166
Rapid adoption20% prior
112.5721
Jevons expansion15% prior
112.2330
Physical bottleneck15% prior
102.9375
Slow absorption12% prior
102.7364
External interruption8% prior
100.1346
Company attribution · 36 months
Technical promise is not the same as realizable gain
Financing runway and timing act as explicit gates—not hidden judgment.
ArchetypeTechnical potentialRealized gain
Enterprise AI adoption12.13%12.13%
Cloud & data centers7.20%7.20%
Compute platforms3.50%3.50%
Capital-constrained innovator9-month runway / 24 months required10.00%3.75%financing-gated
Networking platforms2.60%2.60%
Power equipment2.20%2.20%
The system beneath the number
Explainable by construction
Meridian models the economic domain. AIPI publishes the scalar. M1 tests it against reality.
01
Meridian
A directed graph represents companies, technologies, energy, finance, and their timed relationships.
Read the Meridian explanation 02AIPI
Declared paths, weights, gates, and scenarios resolve into a multi-horizon productivity index.
Read the AIPI explanation 03M1 ledger
Pre-registered forecasts score the methodology prospectively, creating a self-calibrating evidence loop.
Read the M1 explanationDeterministic replaySame frozen input produces the same release hash.
Fail-closed validationMalformed probabilities, weights, and references are rejected.
Branch preservationUnresolved coordinates remain explicit instead of disappearing.
01
The map
What Meridian does
Meridian is the explanatory machinery beneath AIPI. It represents the AI economy as a directed graph: companies, technical capabilities, compute, networking, data centers, energy, financing, and adoption are cells; declared relationships carry effects between them with direction, strength, delay, and scenario sensitivity.
- Inputs: time-stamped evidence, declared assumptions, scenarios, company weights, and financing runway.
- Computation: effects travel only along explicit paths and arrive only after declared lags.
- Output: attributable productivity effects by company, relationship, scenario, and horizon.
In plain English: Meridian is the map that explains where productivity might come from and what could prevent it from arriving.
02
The number
What AIPI measures
AIPI is the published index derived from that map. Based at 100, it expresses the probability-weighted expected change in AI-enabled productive capacity—not AI stock performance, investment spending, or sentiment—at several future horizons.
- Term structure: separate 6-, 12-, and 36-month values show when gains may become economically available.
- Scenario field: the central value never erases the named futures underneath it.
- Evidence state: coverage and confidence are published beside the number, but cannot secretly alter it.
In plain English: AIPI is the number that summarizes expected AI productivity while keeping its timing, uncertainty, and evidence quality visible.
03
The reality check
What the M1 ledger does
M1 is the prospective evidence loop. Before an event occurs, the system records the event universe, objective resolution rule, probability, model version, and Git receipt. After resolution, software computes the score without choosing only favorable examples.
- Pre-registration: predictions and scoring rules are sealed before outcomes are known.
- No cherry-picking: a fixed inclusion protocol determines which events enter the ledger.
- Calibration: forecast errors become evidence for revising Meridian’s declared weights and assumptions.
In plain English: M1 tells us, over time, whether AIPI’s machinery deserves trust and where it needs correction.
S
The integrity guard
What SARAI protects
SARAI tests whether governing intent survives as structured reasoning moves between AI systems. In this MVP it detects authorization drift and silent erasure of planted ambiguity using ground truth constructed in advance and deterministic scoring.
- Transport: did the next system reconstruct the declared state correctly?
- Collapse: did an unresolved branch disappear or become falsely certain?
- Authorization: did scope expand, a prohibition vanish, or delegation exceed its limit?
In plain English: SARAI prevents the analytical chain from becoming more certain—or more authorized—than the evidence permits.
Release receipt
sha256:36b0bdc110733236083c0dccf8743b732b4730064ad6b93db65f275f1f4625f1