Forecasting AI-Era Productivity: The Intellectually Converged Human Framework and a Missing Cognitive Mediator in Production Function Theory

Why does massive AI investment fail to generate commensurate productivity gains? We argue the paradox is theoretically generated: prevailing production function frameworks encounter a structural boundary by treating AI as a separable factor of production without modeling the cognitive mediation through which AI generates productive value. This directs investment toward deployment when productivity requires prior development of what we term convergence capacity (C). We propose the Intellectually Converged Human (ICH) framework, a fifth-stage framework for production function theory: H-hat = H[1 + phi(A,C)], where effective productive capacity equals human capital (H) scaled by an augmentation factor [1 + phi], with phi jointly determined by AI utilization intensity (A) and convergence capacity (C), a four-dimensional cognitive construct encompassing embodied understanding, metacognition, temporal integration, and integrative thinking. The production function Y = F(K, H-hat) provides a human-centered mechanism for Solow's TFP residual: A_Solow = [1 + phi(A,C)]^(1-alpha). The framework predicts three augmentation regimes with distinct policy implications. Descriptive cross-national analysis of 20 OECD economies shows the AIxC interaction is associated with 86% of TFP variance versus 31% for AI alone, a pattern-consistent finding in the small-n theoretical tradition. South Korea exemplifies national-scale under-augmentation: high H, substantial A, low C produce phi = 0. We distinguish convergence capacity from adjacent constructs, absorptive capacity, dynamic capability, and human capital, and demonstrate that C constitutes the specific cognitive mediator that prior frameworks have left implicit. We derive C-first policy prescriptions and offer three empirically testable propositions with a falsifiable 10-year forecast.

Summary

Main Finding

The paper proposes the Intellectually Converged Human (ICH) framework: AI is not an independent production factor but a human-augmentation input whose productive effect is jointly determined by AI utilization intensity (A) and a distinct cognitive mediator—convergence capacity (C). Formally: - Ĥ = H · [1 + φ(A, C)] - Y = F(K, Ĥ) In a Cobb–Douglas form this implies Solow TFP is endogenized as - A_Solow = [1 + φ(A, C)]^{1−α}. Empirically, across 20 OECD countries the interaction AI × C explains far more TFP variance (R^2 ≈ 0.86) than AI alone (R^2 ≈ 0.31), suggesting that failure to build convergence capacity explains why large AI investments often fail to produce commensurate productivity gains (the “AI productivity paradox”).

Key Points

Conceptual innovation: Replaces treating AI as a separable factor with an augmentation function φ(A, C) that scales human productive capacity H.
Convergence capacity (C): a four-dimensional cognitive construct essential to converting AI outputs into productive value:
Embodied understanding (domain-grounded skills),
Metacognition (calibration, judgment of AI outputs),
Temporal integration (long-horizon reasoning and sequencing),
Integrative thinking (synthesizing across domains and representations).
Distinction from adjacent concepts: C is argued to be distinct from absorptive capacity, dynamic capabilities, and standard measures of human capital/cognitive skill; it is the specific cognitive mediator for human-AI augmentation.
Architectural/ontological argument: Current AI architectures (e.g., Transformer-based LLMs) are fundamentally “averaging” or predictive devices that cannot, by themselves, restructure problems, ground outputs in embodied context, or perform the meta-judgment necessary for many productive tasks—hence the need for convergent human cognition.
Augmentation regimes: The framework defines three augmentation regimes (one labeled explicitly as Regime I — under-augmentation) in which φ behaves differently; these regimes imply different investment sequencing and policy responses (paper emphasizes a “C-first” prescription).
Empirical pattern (descriptive): Cross-national OLS on 20 OECD economies shows AI adoption alone explains ~31% of TFP variance (p = .011); including AI × C interaction raises explained variance to ~86% (pattern-consistent evidence in small-n tradition). South Korea presented as a deviant case: high H and A but low C and low TFP growth (~0.48% annually).
Forecasting and testability: The paper offers three falsifiable propositions and a 10-year forecast linking C development to realized productivity gains.

Data & Methods

Methodological approach: Conceptual/theoretical framework-building complemented with empirical grounding in a small-n comparative tradition (pattern-consistent illustration rather than causal identification).
Literature: Systematic search (Web of Science, Scopus) using production function, AI, human capital, augmentation, convergence capacity, metacognition terms; identified 97 sources (87 cited).
Empirical data sources: OECD (2024a, 2024b), McKinsey Global Institute (2023), World Bank (2024); constructed dataset for 20 OECD economies.
Analysis: Six OLS models presented as descriptive; key finding is the large increase in explained TFP variance when including AI × C. Devian case analysis applied to South Korea. Code and constructed dataset are stated to be available on request.
Limitations acknowledged: Small-n cross-national design; OLS results are descriptive and pattern-consistent—causal identification deferred to future work with exogenous variation strategies.

Implications for AI Economics

Theoretical: Provides an explicit mechanism linking AI deployment to Solow TFP via human cognitive mediation (φ(A, C)), shifting theoretical emphasis from AI-as-factor to AI-as-augmenter of human capacity.
Measurement: Calls for operationalizing and measuring convergence capacity (C) separately from traditional human capital indicators—moving beyond years of schooling or generic cognitive-skill scores to measures of metacognition, integrative problem solving, domain grounding, and temporal reasoning.
Policy sequencing: Argues for “C-first” strategies—invest in developing convergence capacity (education, training, organizational capability, institutional complements) before or alongside large-scale AI deployment. AI-first investment sequencing risks under-augmentation and wasted deployment dollars.
Organizational strategy: Firms should prioritize building human-AI workflows, metacognitive training, domain grounding processes, and integrative teams that can calibrate, validate, and extend AI outputs rather than only acquiring AI tools.
Forecasting & evaluation: Forecasts of AI-driven growth must be conditional on C development; cross-national and firm-level forecasts should include interaction terms (AI × C) to avoid overoptimistic estimates.
Research agenda: Empirically testable directions include (1) precise operationalization of C and its subcomponents, (2) causal identification of φ via exogenous shocks or randomized interventions that raise C, and (3) longitudinal evaluation of the paper’s 10-year falsifiable forecast.

Caveats: The empirical evidence presented is descriptive and based on a small-n cross-national approach; causal claims are tentative. The framework is intended as an extension to, not a replacement of, task-based models—adding a human-cognitive mediator to explain variation in productivity outcomes from AI deployment.

Assessment

Paper Typetheoretical Evidence Strengthmedium — The paper advances a clear, falsifiable theoretical mechanism (the ICH framework) which gives a plausible explanation for weak productivity responses to AI; empirical support is limited to small-n, cross-national correlations that are suggestive but not causal, so the overall evidentiary package is stronger than pure theory but falls short of robust causal proof. Methods Rigormedium — Theoretical development appears careful and internally consistent (formalizing phi(A,C) and deriving implications for Solow residuals), and the empirical exercise uses an interaction specification that directly tests the theory; however, the empirical sample is small (20 countries), measurement of the novel C construct is likely ad hoc and unvalidated, and there is no strategy to address endogeneity, omitted variables, or measurement error. SampleDescriptive cross-national sample of 20 OECD economies; outcome is national TFP (Solow residual); key predictors are an AI intensity/investment measure (A) and a constructed multi-dimensional convergence-capacity index (C) capturing embodied understanding, metacognition, temporal integration, and integrative thinking; analysis appears to be cross-sectional (or short panel) and reports R-squared comparisons for AI alone versus AI*C interaction. Themesproductivity human_ai_collab skills_training adoption IdentificationProposes a structural theoretical identification by modifying the production function to include an augmentation factor phi(A,C) that scales human capital; empirically tests the theory with descriptive cross-national regressions of TFP on AI intensity (A), a constructed convergence-capacity index (C), and their interaction (A*C) across 20 OECD countries (associational inference only; no exogenous variation or causal identification strategy). GeneralizabilitySmall-n OECD sample limits statistical power and external validity, Novel construct C lacks established measurement/validation, raising measurement-error concerns, Cross-sectional/associational design prevents causal generalization (reverse causality, omitted variables), National-level aggregates mask firm- and sector-level heterogeneity in AI adoption and human capital complementarities, Findings may not apply to non-OECD or lower-income countries with different institutions and labor market structures, Cultural, institutional, and policy differences across countries may confound observed AI×C patterns, Time horizon: short-term cross-sectional patterns may not hold over the long transitional dynamics of AI diffusion

Claims (10)

Claim	Direction	Outcome	Confidence & Evidence	Details
Massive AI investment has failed to generate commensurate productivity gains (the "AI productivity paradox"). Firm Productivity	negative	productivity gains (total factor productivity / output per worker)	Reading fidelity high Study strength medium	not reported 0.12
Prevailing production-function frameworks encounter a structural boundary because they treat AI as a separable factor of production without modeling the cognitive mediation through which AI generates productive value. Organizational Efficiency	negative	adequacy of production-function frameworks to capture AI-driven productivity	Reading fidelity high Study strength speculative	not reported 0.02
Investment is being directed toward AI deployment when achieving productivity gains requires prior development of convergence capacity (C), leading to a misallocation of investment. Task Allocation	negative	alignment of AI investment with productivity-enhancing prerequisites (convergence capacity)	Reading fidelity high Study strength speculative	not reported 0.02
The paper proposes the Intellectually Converged Human (ICH) framework with H-hat = H[1 + phi(A,C)], where effective productive capacity equals human capital (H) scaled by augmentation factor [1 + phi], and phi is jointly determined by AI utilization intensity (A) and convergence capacity (C). Organizational Efficiency	positive	effective productive capacity (H-hat) as a function of human capital and augmentation factor	Reading fidelity high Study strength speculative	H-hat = H[1 + phi(A,C)] 0.02
Using H-hat in the production function Y = F(K, H-hat) provides a human-centered mechanism for Solow's TFP residual: A_Solow = [1 + phi(A,C)]^(1-alpha). Firm Productivity	positive	Solow total factor productivity residual (A_Solow)	Reading fidelity high Study strength speculative	A_Solow = [1 + phi(A,C)]^(1-alpha) 0.02
The ICH framework predicts three distinct augmentation regimes (determined by combinations of A and C) with distinct policy implications. Organizational Efficiency	mixed	augmentation regime classification (regimes of phi behavior as functions of A and C)	Reading fidelity high Study strength speculative	not reported 0.02
A descriptive cross-national analysis of 20 OECD economies shows the AI × C interaction is associated with 86% of TFP variance, versus 31% for AI alone. Firm Productivity	positive	proportion of variance in total factor productivity (TFP) explained	Reading fidelity high Study strength medium	n=20 86% of TFP variance (AI × C interaction); 31% of TFP variance (AI alone) 0.12
South Korea exemplifies national-scale under-augmentation: high human capital (H), substantial AI (A), but low convergence capacity (C) produce phi = 0. Firm Productivity	null_result	augmentation factor phi (measured as zero for the South Korea example)	Reading fidelity high Study strength low	n=1 phi = 0 0.06
Convergence capacity (C) is distinct from absorptive capacity, dynamic capability, and human capital, and constitutes the specific cognitive mediator prior frameworks have left implicit. Skill Acquisition	positive	theoretical distinctness and mediating role of convergence capacity (C)	Reading fidelity high Study strength medium	not reported 0.12
The paper derives C‑first policy prescriptions and offers three empirically testable propositions along with a falsifiable 10-year forecast. Governance And Regulation	positive	policy prescriptions and testable empirical propositions (with 10-year forecast horizon)	Reading fidelity high Study strength speculative	not reported 0.02

AI spending often fails to boost output because nations lack the cognitive 'convergence capacity' needed to translate tools into productive work; a cross-country interaction between AI intensity and a four‑part convergence index explains far more TFP variation than AI alone.