A review of 138 papers finds generative AI is reshaping organizational strategy and operations and can drive sustainable digital transformation that improves firm performance — but gains depend on strong ethical governance and effective human–AI collaboration.

Main Finding

Generative AI (GenAI) is a distinct strategic capability (not just a generic “AI” tool) that can drive sustainable digital transformation and thereby improve multidimensional organizational performance. The authors’ systematic review of 138 peer-reviewed articles (2020–2025) identifies six thematic channels through which GenAI operates, and proposes an integrative conceptual model: GenAI adoption → sustainable digital transformation (mediator) → organizational performance, with ethical governance and human–AI collaboration as key moderators.

Key Points

• Scope and contribution

  • Focuses on business/management literature (Scopus & Web of Science) between 2020–2025.
  • Frames GenAI as qualitatively different from prior AI: creates content/insights and reshapes strategy/innovation, not only automation.
  • Integrates sustainability (ESG) into the GenAI–digital transformation–performance nexus.

• Six themes from thematic analysis

• Strategic integration of GenAI: GenAI as an organizational resource enabling new business models, quicker innovation cycles, and strategic forecasting.
• Operational efficiency & process automation: Automates knowledge-intensive tasks, speeds operations, reduces errors and costs, and allows redeployment of human labor to higher-value tasks.
• Data‑intelligent, data‑driven decision-making: Handles large structured/unstructured data, scenario modeling and predictive simulation to support evidence-based management.
• Human–AI collaboration: GenAI augments human work (especially creative and decision tasks); success requires training, culture, and new skills.
• Ethics, governance & risk: Issues include privacy, bias, transparency and accountability; governance frameworks are necessary to maintain trust and compliance.

• Sustainability & performance outcomes: GenAI can support environmental and social goals (energy optimization, emissions reduction, ESG reporting) while improving financial, operational, and innovation performance.

• Conceptual model

  • Sustainable digital transformation mediates the effect of GenAI adoption on organizational performance.
  • Ethical governance and human–AI collaboration moderate the magnitude and sustainability of performance gains.

• Limitations highlighted by authors

  • Literature-based synthesis (no new empirical estimation).
  • Search limited to English-language, peer-reviewed business/management articles; excludes technical/engineering-only studies and non-journal outputs.
  • Time window focused on recent GenAI diffusion (2020–2025).

Data & Methods

• Method: Systematic Literature Review (SLR) following PRISMA guidelines.
• Databases: Scopus and Web of Science.
• Query (conceptual): (“Generative AI” OR “Artificial Intelligence”) AND (“Digital Transformation” OR “Business Transformation”) AND (“Sustainability” OR “Organizational Performance”).
• Time window: publications from 2020 to 2025 (English only).
• Inclusion: peer‑reviewed journal articles in business/management focusing on GenAI or advanced AI applications with organizational relevance (empirical, conceptual, or review).
• Exclusion: conference papers, editorials, book chapters, engineering/technical-only articles.
• Final sample: 138 articles after duplicates and eligibility screening.
• Analysis: Iterative thematic coding and synthesis to extract recurrent themes and relationships; development of an integrative conceptual framework.

Implications for AI Economics

• Productivity and TFP

  • GenAI may raise firm-level productivity through automation of knowledge work, improved decision-making, and faster innovation cycles. Economists should estimate GenAI’s contribution to TFP and decompose sources (automation vs. complementary skill effects).
  • Empirical approaches: firm-level panel DiD exploiting staggered GenAI tool rollouts, synthetic control for adopters, or instrumental variables capturing exogenous exposure (e.g., cloud availability, vendor rollout geography).

• Labor markets and skill composition

  • Expect reallocation from routine tasks toward higher-skill activities; potential increase in demand for cognitive, analytic, and AI-complementary skills.
  • Research directions: estimate effects on wages, employment composition, hours, and returns to skills (difference-in-differences by occupation/task exposure). Examine heterogeneous effects across firm size and sectors.

• Innovation, R&D and competition

  • GenAI can shorten innovation cycles and lower marginal costs of ideation, potentially increasing product variety and R&D productivity.
  • Important to study impacts on market structure: does GenAI favor incumbents (scale & data advantages) or lower entry costs for startups? Use patent data, product introductions, and market concentration measures over time.

• Returns to capital and investment behavior

  • GenAI investments may act as intangible capital with long-run productivity returns. Measure CAPEX vs. intangible investment, adoption lags, and persistence of returns.
  • Micro-econometric estimation of investment-to-output elasticities and adjustment costs is needed.

• Distributional and welfare effects

  • Evaluate who captures gains: firms, workers, consumers. Consider consumer surplus from improved products vs. redistribution from displaced workers.
  • Policy-relevant welfare analyses should incorporate externalities (privacy, bias), and environmental impacts (net energy use).

• Sustainability and externalities

  • GenAI can reduce resource use via optimization but also impose energy/computation costs. Measure net environmental impact (life-cycle analysis, firm-level energy use pre/post adoption).
  • ESG-aligned GenAI adoption could change firm valuations and investor behavior—study asset prices and cost of capital differentials for adopters.

• Governance, regulation, and adoption frictions

  • Ethical governance moderates realized gains; weak governance can produce reputational and regulatory costs that reduce net benefits.
  • Study how regulatory regimes and internal governance shape adoption decisions and returns—use cross-jurisdiction variation and natural experiments (policy changes).

• Measurement recommendations for empirical work

  • Construct measures of GenAI adoption intensity (software subscriptions, API usage, job postings requiring GenAI skills, text/code generation logs).
  • Create task-exposure indices using occupational-task content to predict which jobs/firms are most affected.
  • Combine administrative microdata (employment, payroll) with firm financials and product/innovation outcomes.

• Policy implications

  • Support worker retraining and complementary skill development to capture productivity gains equitably.
  • Encourage governance standards and transparency to limit negative externalities and preserve trust.
  • Consider incentives for low-carbon AI deployment (efficiency standards, green compute subsidies) if computation carbon intensity is nontrivial.

• Suggested empirical agendas

  • Causal impact studies: DiD, event studies around major GenAI tool releases, randomized trials for adoption support programs.
  • Longitudinal studies of firm-level performance and labor outcomes across sectors.
  • Macro modeling: incorporate GenAI as a general-purpose technology in growth models to assess long-run aggregate effects and transitional dynamics.

If you’d like, I can draft a short list of empirical designs (data sources and identification strategies) tailored to a specific question (e.g., estimating GenAI’s effect on firm productivity or on occupational wages).