A unified framework places algorithmic prediction, human judgment and ethical governance at the centre of AI-driven economic decision-making; it argues that trustworthy AI, explainability and human oversight are essential for sustainable decisions but provides no empirical validation.

Main Finding

The paper develops the AI-EDG Framework (Artificial Intelligence‑Driven Economic Decision and Governance Framework), a conceptual, end‑to‑end model that reconceptualizes economic decision‑making as a cyclical socio‑technical system. Decision quality is framed as an emergent outcome of the integrated interaction among (1) a data ecosystem, (2) AI analytics, (3) human–AI collaborative reasoning, (4) ethical governance, (5) decision implementation, and (6) continuous learning/feedback. Trustworthy governance (explainability, human oversight, data quality, accountability, institutional compliance) is argued to be as important as predictive performance for sustainable AI‑driven economic decisions.

Key Points

• Gap addressed: existing literature treats technical performance, human–AI collaboration, and governance largely separately; AI‑EDG integrates these into a unified decision cycle spanning individuals, firms, public institutions, and labor markets.
• Core components of AI‑EDG: data ecosystem → AI analytical core → human–AI collaborative reasoning → ethical governance layer → decision implementation → continuous feedback/learning loops.
• Conceptual shift: from automation/intuition-driven decisions to a division of labor where AI reduces the cost of prediction while human agents assign value, set goals, and provide ethical/judgmental oversight.
• Differences between traditional and AI‑driven decision‑making emphasized:
  • Descriptive → predictive & prescriptive decision support
  • Intuition → data‑driven architectures (complementary roles for System 1/System 2)
  • Blurred separation of prediction (machine) and judgment (human)
  • Managers’ role evolves into “decision architects” who design data, models, oversight, and governance
  • From pure automation to human‑supervised socio‑technical systems where explainability, trust, and institutional legitimacy matter
• Policy and governance priorities identified: trustworthy AI governance, explainability (XAI), human oversight, data quality management, accountability, regulatory compliance, and labor market policies to support skill adaptation and pro‑worker AI design.
• Proposition: quality of AI‑driven economic decision‑making is emergent from the integrated interaction among high‑quality data, AI analytics, human judgment, and ethical governance.
• Contribution: interdisciplinary conceptual integration (economics of prediction, decision theory, DSS, XAI, AI governance) and a testable theoretical scaffold for empirical research and policy design.

Data & Methods

• Methodological approach: conceptual synthesis / systematic review of interdisciplinary literature spanning economics, decision sciences, artificial intelligence, information systems, and public policy.
• Sources and grounding: draws on foundational works (e.g., Simon on bounded rationality; Kahneman on System 1/2; Agrawal et al. on prediction machines), recent governance frameworks (OECD, NIST), and applied AI/DSS literature.
• No primary empirical data or quantitative tests are presented—AI‑EDG is a theoretical, integrative framework intended to guide empirical work and policy evaluation.
• Limitations noted implicitly: framework is conceptual and requires operationalization and empirical validation across sectors and scales.

Implications for AI Economics

• Theoretical implications:
  • Reframes AI’s economic role from mere productivity/automation effects to capability augmentation that changes informational foundations of decisions and institutional responsibilities.
  • Suggests new outcome constructs beyond accuracy (trustworthiness, explainability, institutional legitimacy, distributional impacts) that economists should incorporate in models of AI adoption and welfare analysis.
  • Motivates formal models distinguishing prediction (machine) and valuation/judgment (human) and analyzing complementarities (e.g., skill‑biased augmentation vs. displacement).
• Empirical research agenda:
  • Operationalize AI‑EDG components (data quality, model explainability, oversight intensity, governance compliance) and estimate their causal effects on organizational/policy outcomes.
  • Sectoral empirical studies (public sector procurement/policy, financial services, labor markets) and comparative institutional analyses.
  • Longitudinal and experimental designs to trace feedback loops and learning dynamics (how decisions informed by AI change data ecosystems and institutions over time).
  • Measurement development: composite indices for decision quality that include trust, fairness, and accountability metrics, not only prediction error.
• Policy and regulatory implications:
  • Regulatory design should mandate and incentivize explainability, human‑in‑the‑loop oversight, data quality governance, and accountability mechanisms (audit trails, impact assessments).
  • Labor policy: support re‑skilling and role redesign so AI acts as pro‑worker augmentation where possible; anticipate distributional effects and design transition supports.
  • Public sector: institutionalize AI governance standards (aligned with OECD/NIST principles) across procurement, transparency, and democratic accountability.
• Practical implications for firms and institutions:
  • Treat managers as decision architects — invest in data infrastructure, model governance, XAI tools, oversight protocols and feedback systems.
  • Evaluate AI deployment on multi‑dimensional criteria (efficiency, explainability, legal/ethical compliance, societal legitimacy).
• Overall: economists studying AI should integrate governance variables into models of adoption and welfare; policymakers should assess AI systems holistically—technical performance plus institutional and ethical dimensions—to ensure socially beneficial outcomes.