AI has graduated from back-office automation to a strategic organisational capability, transforming decision-making and human–machine collaboration; however, entrenched problems—algorithmic bias, opaque models, shaky data governance and workforce adjustment—threaten to blunt its productivity and competitive gains.

Main Finding

AI has evolved from a narrow operational tool into a strategic, organisational capability that reshapes managerial decision-making, organisational design and firm-level value creation. However, literature is fragmented, often technology-centric, and highlights persistent challenges—algorithmic bias, opacity, data governance, regulatory uncertainty and workforce adaptation—that condition the realised economic effects of AI.

Key Points

• Conceptual framing

  • AI is best understood as a socio-technical and managerial phenomenon: outcomes depend on human–AI collaboration, governance, and strategy, not just on algorithmic performance.
  • Recent advances (ML, deep learning, LLMs/generative AI) extend AI from analytics to creative/strategic tasks.

• Organisational applications (representative firm examples)

  • Operations & supply chains: Amazon, Volkswagen, Walmart and Alibaba Cloud report measurable gains (reduced delivery costs, fewer stoppages, lower inventory costs; cited figures: VW production stoppages −15%, energy costs −12%; Walmart stock-outs −18%, ≈USD 200m annual savings).
  • Finance & risk: JPMorgan reduced credit approval times dramatically and increased approvals; Morgan Stanley, Goldman Sachs and IBM cases show faster advisory, reporting and fraud detection.
  • R&D & innovation: GE and Moderna used generative AI to compress innovation cycles and cut R&D costs (Moderna reported faster candidate prep and ≈USD 150m R&D savings).
  • Customer service: Microsoft Copilot and Tencent chatbots improved worker productivity and customer-response times.

• Management research themes

  • Strategic management: AI enhances data-driven strategic foresight but risks over-reliance, bias and opacity; hybrid human–machine governance is crucial.
  • HRM: AI transforms recruitment, evaluation and HR roles; raises fairness, trust and inequality concerns; notion of “Human–Technology Resource Management.”
  • Marketing: powerful for personalisation and analytics but creates privacy, trust and regulatory risks.
  • Governance & ethics: growing literature on accountability, data governance, explainability; consensus on principles (transparency, fairness, accountability, privacy) but operationalisation remains contested.

• Gaps and limits

  • Fragmented literature across domains; predominance of descriptive/technology-centric studies.
  • Need for causal, firm-level evidence on economic impacts, distributional effects, and long-run organisational change.

Data & Methods

• Study type: integrative literature review and thematic synthesis of international studies in business, management and related fields.
• Sources: empirical case studies, industry reports and academic literature spanning operations, finance, marketing, HRM, strategy and governance.
• Methods used in reviewed studies (summary):
  • Case-based evidence (firm reports, corporate collaborations).
  • Empirical analyses in management/IS using surveys, firm-level outcomes, and observational data.
  • Conceptual and normative frameworks for governance and ethics.
• Notable empirical evidence cited: quantitative firm-level outcome metrics from corporate disclosures and industry reports (e.g., % reductions in stoppages, time savings, dollar cost savings).
• Limitations of the review method:
  • Relies on published and corporate-reported outcomes (potential selection/publicity bias).
  • Heterogeneous methodologies in underlying studies restrict cross-study causal inference.

Implications for AI Economics

• Productivity and growth
  • AI is a distinct form of capital that can raise firm productivity, shorten innovation cycles and lower costs. Accurate measurement of AI capital and its depreciation is needed for growth accounting.
• Firm heterogeneity and market structure
  • AI can amplify returns to scale and data advantages—potentially increasing concentration and incumbency advantages. Empirical IO work should quantify AI’s role in markups, entry barriers and market power.
• Labor markets and distribution
  • AI reshapes task content: automating routine tasks while complementing high-skill activities. Research should quantify job reallocation, wage dispersion, and skill-biased complementarities across sectors and demographics.
• Investment, adoption and complementarities
  • Returns to AI depend on complementary investments (data infrastructure, human capital, governance). Estimating complementarities and second-order adoption effects is essential for policy and firm strategy.
• Externalities and public goods
  • Data externalities (network effects, non-rival data), algorithmic spillovers and systemic risk (e.g., correlated model failures) imply potentially large social externalities—motivating regulation, data-sharing frameworks and coordination policies.
• Regulation and policy
  • Regulatory uncertainty affects adoption timing and competitive dynamics. Economic research should evaluate trade-offs of antitrust, data governance, liability rules and retraining subsidies.
• Measurement and empirical strategies (recommended)
  • Develop firm-level measures of AI capital: expenditures, cloud/compute usage, models deployed, patents, and job-posting skill content.
  • Identification strategies: difference-in-differences on staggered AI rollouts, instrumental variables (e.g., exogenous compute or data shocks), event studies around AI product launches, regression discontinuity (procurement thresholds), and matched firm comparisons.
  • Use administrative data, matched employer–employee datasets, online job postings, and digital trace data (API/model logs) to link AI adoption to productivity, wages, and market outcomes.
  • Structural and general-equilibrium models to assess long-run distributional and macro effects, as well as optimal policy design.
• Priority research questions for AI economics
  • What are the causal returns to AI investment at the firm and plant level, and how do they vary by industry and firm size?
  • How does AI adoption affect market concentration, entry, and consumer welfare?
  • How do AI and human capital interact—when does AI substitute vs. complement labor—and what are the wage and employment implications across skill groups?
  • How do governance regimes (data access, liability, transparency requirements) alter incentives to invest in and deploy AI?
  • What are the economy-wide externalities (network, informational, systemic risk), and what policy instruments internalise them effectively?

Overall, the reviewed literature establishes AI as an economically consequential and institutionally embedded technology. AI economics should move from descriptive case studies toward rigorous causal and structural analyses that quantify productivity, distributional and market-structure effects, and that inform policy design for inclusive, competitive and well-governed AI-driven growth.