When will AI measurably change economic growth?

Economic adoption is not economic causation

Companies can invest heavily in AI, deploy new tools and report employee time savings without generating a proportional macroeconomic productivity gain. An increase in usage is not the same outcome as higher output per hour of work. Cortist tracks the difference between adoption, investment, firm-level experiments and national productivity statistics.

OECD evidence on adoption

An OECD 2026 study covering 15 countries finds that larger firms tend to adopt advanced technologies more widely and that skill and digital infrastructure differences help explain productivity premia. This cautions against treating the observed success of adopters as the causal impact of AI alone. Sector and firm comparisons must control for selection.

What the IMF and OECD reports indicate

The IMF discussed AI-related technology investment as a contributor to recent US growth but described productivity gains as still unfolding and partly uncertain. OECD indicators show US labor-productivity acceleration but warn about causal interpretation. Neither report alone establishes a one-percentage-point AI-caused gain across a shared three-year window.

Why we need two independent studies

Productivity is affected by business cycles, ordinary capital deepening, wages, regulation, demographic change, training and measurement practices. A single estimated AI contribution could rest on unverified assumptions. The proposed F-007 test therefore calls for independent analyses using a documented no-AI counterfactual and comparable observed years.

What is a percentage-point increase?

If labor productivity would have grown 1.5% per year without AI, an additional one percentage point would mean an AI-attributed growth rate around 2.5% annually. That example is explanatory, not a prediction of the actual series. The outcome refers to additional annual growth rather than a one-off level gain or a company revenue multiple.

What accelerates a durable payoff?

Successful organizational integration, worker training, complementary software, reliable automation and competition could turn potential time savings into output gains. Bottlenecks include poor work redesign, integration costs, trust problems, energy constraints and uneven diffusion. The Jobs forecast concerns hiring; it cannot substitute for measured macro productivity.

Why Cortist starts at 35%

The draft F-007 estimate assigns 35% by 2035 to this stringent, causally attributable outcome. This is a subjective judgment; the interpolated year-slider values are assumptions, not official economic forecasts. If the data begin to support a stronger conclusion, we will publish a new version alongside the older estimate.

Difficulties in adjudication

Independent reviewers still need to agree on acceptable counterfactual identification, confidence levels and whether the studies truly share an observed three-year period. It may prove impossible to settle definitively without new measurement practices. If it cannot be scored fairly, the model must keep a HOLD or unresolved status rather than pretend certainty.

How evidence will change this forecast

Cortist monitors original datasets, official measurements and independent evaluations rather than treating manufacturer, researcher or policymaker claims as settled outcomes. Any material reassessment must disclose the previous value, the new evidence, counterarguments and the reason for changing the subjective assessment. The older snapshot will remain at its original versioned URL.

Before this question becomes eligible for formal scorekeeping, independent reviewers must agree on the boundary of the event, adjudication rules, data sources and any necessary fallback measurement. If an essential definition changes, a fresh forecast question will be published instead of rewriting an old one.