← Papers

Digital tools can accelerate green innovation and cut emissions—but only with clean power, enforceable auditability and upgraded workforce skills; absent these complements, AI-driven digitalization risks higher electricity use, measurement divergence and widening inequality.

The synergy of digital innovation and green economy: A systematic review of mechanisms, challenges, and adaptive strategies in the post-Al era

R. Santos, María Isabel Casal Reyes · Fetched March 15, 2026 · Chinese Studies Monthly

semantic_scholar review_meta medium evidence 7/10 relevance DOI Source PDF

Digitalization—notably AI, analytics, and platform infrastructure—can speed green innovation and reduce emissions, but these benefits materialize only when combined with credible standards/auditability, low-carbon power, and workforce capability building; otherwise digital expansion can increase electricity demand, measurement divergence, greenwashing, and inequality.

Digital transformation and the green transition increasingly co-evolve through shared infrastructures (data, energy, and institutions), yet their interaction is not automatically synergistic. This review synthesizes peer-reviewed research and authoritative institutional reports to clarify how digital technologies—especially AI, analytics, and platform infrastructures—shape green economic outcomes, and how environmental constraints and governance feedback reshape digital diffusion. We organize evidence around four domains: (i) labor-market restructuring under AI and digitalization, with attention to wage polarization, rents, and institutional mediators; (ii) skills mismatch and SME adoption constraints as a binding bottleneck for inclusive digital-green upgrading; (iii) the convergence of green finance and computing, where automated ESG assessment expands monitoring capacity but also amplifies measurement divergence and greenwashing risks; and (iv) sustainable infrastructure and energy transition, focusing on hydrogen value chains and the energy footprint of digital systems (data centers and AI workloads). A sectoral case—digital tourism—illustrates both substitution potential (virtual experiences, demand management) and rebound risks. Evidence converges on a conditional-synergy thesis: digital tools can accelerate green innovation and emissions reductions when coupled with credible standards, auditability, clean power, and workforce capability building; absent these complements, digitalization may increase electricity demand, widen inequality, and incentivize strategic disclosure. The review identifies three research gaps that limit policy inference: long-horizon causal evidence on non-linear coupling between digitalization and decarbonization, joint modeling of distributional outcomes and environmental performance, and integrated evaluation of AI-enabled sustainable finance under heterogeneous disclosure regimes. We propose a future agenda that prioritizes enforceable AI governance, life-cycle carbon accounting across hydrogen supply chains, and targeted SME capability policies.

Summary

Main Finding

Digital innovation (AI, analytics, platforms) can accelerate green innovation and emissions reductions, but synergy with the green economy is conditional — not automatic. Positive environmental outcomes require complementary institutions and inputs (credible standards and enforcement, auditability, clean power, workforce capability). Without these complements, digitalization can widen inequality, enable greenwashing, and increase electricity demand (including from data centers and AI workloads), shifting rather than resolving environmental pressures.

Key Points

Conditional synergy: Digital tools expand measurement, optimization, and innovation potential, but net-green outcomes depend on governance, energy supply, and distributional policies.
Four concentrated evidence areas:
Labor-market restructuring and inequality — AI induces task substitution and polarization; productivity gains may coincide with stagnant median wages and rising top incomes via wage and rent channels (task-based automation + firm rents).
Skills gaps and SME adoption bottlenecks — SMEs face adoption costs, limited HR capacity, and financing constraints; they may bridge transitions only with modular tools, upskilling, and institutional support.
AI-enabled ESG assessment, measurement divergence, and greenwashing risks — ESG ratings diverge substantially; NLP and remote sensing can scale auditing but need ground-truth, standards, and enforceable liability to avoid false signals and strategic disclosure.
Energy-transition tradeoffs — rising electricity demand from data centers and AI workloads, competition for clean power with electrification and hydrogen electrolysis, and hydrogen supply-chain costs make system-level coordination essential. Rebound effects can erase efficiency gains.
Sector and place matter: built environment, grid reliability, and urban form mediate outcomes; digital tourism exemplifies both substitution potential and rebound/equity risks.
Human capital is central: interdisciplinary, action-oriented education (digital ethics, carbon accounting, governance literacy) is a primary mechanism for translating digital tools into sustainable outcomes.
Key research and policy gaps: long-horizon causal evidence, joint distributional–environmental modeling, evaluation under heterogeneous disclosure regimes, rigorous evaluation of training models, and life-cycle carbon accounting for hydrogen.

Data & Methods

Study type: Systematic structured literature review following PRISMA 2020 principles.
Coverage: Targeted searches of peer‑reviewed articles and authoritative reports, prioritizing 2019–2025 while including foundational theoretical work.
Sources prioritized: academic journals plus institutional reports (OECD, IEA, UNESCO, Equator Principles, EU taxonomy assessments).
Evidence types synthesized: empirical labor- and firm-level studies, sectoral case studies, policy reports, life-cycle and energy-system analyses, NLP/text-as-data studies for greenwashing detection, and required 2025 articles (Gu et al. series) used for mechanism development.
Methodological limits of literature: cross-sectional and short-to-medium horizon studies dominate; causal identification over long horizons and integrated distributional–environmental models are scarce.

Implications for AI Economics

Model distributional impacts jointly with environmental outcomes: AI adoption affects wages, rents, and political economy of decarbonization. Economic models should integrate labor-market polarization, firm market power, and public acceptability of climate policy.
Incorporate energy-grid interactions and rebound effects: quantify electricity demand from AI workloads, data-center trends, and competition for clean power (e.g., hydrogen electrolysis vs. AI loads). Evaluate system-level tradeoffs and opportunity costs of scarce clean electricity.
Evaluate regulation–AI complementarities: AI-enabled monitoring can scale enforcement, but its effectiveness depends on standardized metrics, auditability, and liability. Work on mechanism design and empirical assessment of disclosure regimes (mandatory metrics, taxonomies).
Design SME- and skills‑focused interventions: assess cost‑effective modular tools, shared infrastructure, and upskilling programs; estimate how these reduce inequality and improve diffusion of green technologies.
Improve measurement for green finance: develop and validate ground-truth datasets to train and test NLP and remote-sensing detectors for greenwashing; measure how rating divergence affects capital allocation and firm behavior.
Life-cycle and supply-chain accounting priorities: economics research should adopt full life-cycle carbon accounting (notably for hydrogen) and integrate digital MRV into market design.
Research agenda recommendations:
- Long-run causal studies of AI adoption on emissions and labor outcomes.
- Joint distributional–environmental simulation models to test policy packages.
- Experimental and quasi-experimental evaluations of training/education programs and SME support.
- Empirical testing of AI-based greenwashing detection against verifiable performance data.
- Modeling of electricity-grid capacity expansion scenarios that account for rising digital loads and competing decarbonization demands.

Short takeaway: AI matters for decarbonization but only when paired with enforceable standards, clean-energy expansion, workforce capability building, and policies that address distributional consequences. Economic research should move from “can AI help?” to “under what institutional and infrastructural conditions does AI produce net-social benefits?”

Assessment

Paper Typereview_meta Evidence Strengthmedium — The paper is a systematic synthesis of diverse empirical, modeling, and LCA studies that consistently point to conditional synergies between digitalization/AI and decarbonization; however, the underlying literature is heterogeneous, often short-run or model-based, and lacks long-run causal identification and comparable metrics, limiting confidence in strong causal claims. Methods Rigormedium — The review systematically covers multiple evidence streams (administrative panels, firm surveys, telemetry, LCA, ESG databases, modeling), clearly articulates methods used in the literature and research gaps, and gives concrete empirical and modeling recommendations; but it is a narrative/systematic synthesis rather than a formal meta-analysis, may face selection and publication biases in included studies, and does not itself produce new causal estimates. SampleA systematic review of peer‑reviewed literature and authoritative institutional reports spanning labor-economics panels and microdata, employer and firm surveys, platform and compute telemetry, ESG ratings and disclosure data, life-cycle assessments (notably for hydrogen), sector-specific demand and substitution models, and modeling studies of digital/energy systems. Themesinnovation governance adoption labor_markets skills_training productivity inequality GeneralizabilityFindings aggregate heterogeneous evidence across sectors and countries—effects likely vary by industry, firm size, and national grid carbon intensity, Short time horizons in much of the literature limit inference about long-run, path-dependent outcomes and lock‑ins, Measurement heterogeneity in digital-adoption and ESG metrics reduces comparability across studies and jurisdictions, SME constraints and dynamics differ markedly from large firms, limiting transferability of firm-level results, Regulatory and disclosure regimes vary across countries, so governance-dependent results may not generalize globally, Modeling and LCA results depend on strong technical assumptions (e.g., compute energy per model, hydrogen pathways)

Claims (10)

Claim	Direction	Confidence	Outcome	Details
Digital tools can accelerate green innovation and emissions reductions when coupled with credible standards, auditability, clean power, and workforce capability building. Innovation Output	positive	medium	green innovation activity and greenhouse gas emissions reductions	0.14
Absent complementary institutions and infrastructure, digitalization may increase electricity demand, widen inequality, and incentivize strategic disclosure (greenwashing). Inequality	negative	medium	electricity demand; measures of inequality (e.g., wage distribution); incidence of strategic ESG disclosure/greenwashing	0.14
AI and digitalization are restructuring labor markets, producing wage polarization and rents, with outcomes mediated by labor-market institutions. Wages	mixed	medium	wage distribution/polarization and economic rents captured by workers or firms	0.14
Skills mismatch and SME adoption constraints constitute a binding bottleneck for inclusive digital–green upgrading. Adoption Rate	negative	medium	SME adoption rates of digital/green technologies and inclusiveness of upgrading (distributional access to digital-green benefits)	0.14
The convergence of green finance and computing — especially automated ESG assessment — expands monitoring capacity but also amplifies measurement divergence and greenwashing risks. Regulatory Compliance	mixed	medium	monitoring capacity (coverage/frequency of ESG assessments); measurement divergence across ESG metrics; frequency/evidence of greenwashing	0.14
Sustainable infrastructure and energy-transition analyses must account for hydrogen value chains and the substantial energy footprint of digital systems (data centers and AI workloads). Other	mixed	medium	life-cycle carbon emissions of hydrogen value chains; energy consumption/carbon footprint of data centers and AI workloads	0.14
In digital tourism, there is both substitution potential (virtual experiences, demand management) and rebound risks that may offset emissions reductions. Other	mixed	medium	tourism demand patterns, substitution to virtual experiences, and net emissions from tourism (including rebound effects)	0.14
There is a shortage of long-horizon causal evidence on non-linear coupling between digitalization and decarbonization, limiting robust policy inference. Research Productivity	null_result	high	availability of long-horizon causal studies on digitalization–decarbonization interactions	0.24
Research has insufficiently modeled joint distributional outcomes and environmental performance, and lacks integrated evaluation of AI-enabled sustainable finance under heterogeneous disclosure regimes. Research Productivity	null_result	high	existence of joint models linking distributional (inequality) outcomes and environmental performance; evaluations of AI-enabled sustainable finance across disclosure regimes	0.24
Policy priorities should include enforceable AI governance, life-cycle carbon accounting across hydrogen supply chains, and targeted SME capability policies to realize conditional synergies between digitalization and green transition. Governance And Regulation	positive	medium	adoption/implementation of enforceable AI governance; adoption of life-cycle carbon accounting for hydrogen; improvements in SME digital-green capabilities	0.14