← Papers

AI can cut supply‑chain emissions by improving demand matching, logistics, traceability and supplier collaboration; yet without cross‑firm data governance, explainable models, aligned incentives and life‑cycle carbon accounting those gains may be limited or offset by rebound effects.

Pathways for AI-Driven Green and Low-Carbon Transition of Supply Chains

Jinkai Ke · January 01, 2026 · Modern Economy

openalex theoretical n/a evidence 7/10 relevance Full text usable extracted full text DOI Source PDF

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

OpenAlex

Latest observation: August 25, 2026

Jinkai Ke provider ID

Semantic Scholar

Latest observation: August 25, 2026

Jinkai Ke provider ID

The paper argues AI can enable supply‑chain decarbonization via intelligent demand sensing, logistics scheduling, chain‑wide carbon traceability, and supplier collaboration, but data governance, algorithmic opacity, misaligned incentives, and AI's own energy use pose structural limits requiring institutional and technical reforms.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

0 cumulative citations

OpenAlex · Observed August 25, 2026

View corpus context

0 cumulative citations

Semantic Scholar · Observed August 25, 2026

View corpus context

Global supply chains account for approximately 60% of total carbon emissions worldwide, yet fragmented information across multiple actors and divergent interest objectives render systemic emission reductions unattainable through traditional management approaches. This paper delineates four pathways through which artificial intelligence (AI) drives the green and low-carbon transition of supply chains: intelligent demand sensing enhances supply-demand matching, curbing superfluous emissions at the source; intelligent logistics scheduling optimizes the trade-off among cost, delivery time, and energy consumption; chain-wide carbon footprint traceability addresses Scope 3 emissions; and supplier collaboration and empowerment promote green and low-carbon transformation. It also identifies structural limitations inherent in these pathways, including fragmented data governance, the opacity of algorithmic decision-making rationales, organizational incentive misalignments, and environmental rebound risks arising from the computing power consumption of AI itself. Accordingly, the paper proposes optimization directions such as constructing a cross-organizational carbon data governance framework, developing explainable carbon decision intelligence models, designing collaborative carbon-reduction incentive mechanisms that balance equity and efficiency, and establishing a full-life-cycle carbon performance evaluation system. Embedding institutional design into technical processes enables AI to evolve from an efficiency tool into an institutional infrastructure for the green governance of supply chains.

Summary

Main Finding

AI can enable systemic green and low‑carbon transitions across supply chains via four complementary pathways—intelligent demand sensing, energy‑aware logistics scheduling, chain‑wide carbon traceability, and supplier collaboration—but realizing net environmental and economic benefits requires institutional redesign (cross‑organizational data governance, explainable decision models, incentive‑compatible mechanisms, and full life‑cycle carbon accounting) to address data fragmentation, algorithmic opacity, misaligned incentives, and AI’s own carbon rebound risks.

Key Points

Four AI‑enabled pathways
- Intelligent demand sensing: multi‑source AI forecasting (sales, social media, weather, macro indicators) improves supply–demand matching to reduce overproduction, inventories, and reverse logistics emissions (case: JD Logistics’ MRV‑T).
- Energy‑efficiency logistics scheduling: multi‑objective AI routing/scheduling (traffic, weather, tariffs, node location, multimodal choice) balances cost, time, energy and carbon (case: Schneider Electric).
- Chain‑wide carbon footprint traceability: NLP, computer vision and related methods extract/standardize emissions data from unstructured supplier/transport records, improving Scope 3 traceability.
- Supplier collaboration and empowerment: knowledge graphs and ML supplier scoring (energy intensity, carbon maturity, renewables share) guide green procurement and support SME optimization (case: Alibaba Cloud “Energy Expert”).
Structural limitations
- Fragmented data governance: uneven digitalization, inconsistent formats and weak upstream data availability constrain model accuracy (acute for Scope 3).
- Algorithmic credibility dilemma: opaque models reduce manager trust; explainability is needed for adoption.
- Organizational incentive misalignment: costs of abatement concentrate on focal firms or certain tiers while benefits diffuse—AI alone cannot solve collective action problems and may exacerbate divides.
- Environmental rebound risks: training/inference carbon costs of AI can offset local emission reductions unless accounted for.
Proposed optimizations
- Cross‑organizational carbon data governance, privacy‑preserving collaborative computation, and lightweight cloud tools for SMEs.
- Explainable carbon decision intelligence: prefer transparent models or add post‑hoc explanations; integrate explanations in audit/compliance workflows.
- Incentive mechanisms: AI‑enabled real‑time performance tracking, tiered procurement/payment incentives, and blockchain/smart‑contract automated settlements for carbon credits/transfers to reduce transaction costs and free‑riding.
- Full life‑cycle carbon performance evaluation: include AI training/inference emissions; prioritize lightweight models and green data centers; use net contribution as metric.

Data & Methods

Study type: qualitative, systematic analysis and synthesis.
Sources: recent academic literature on AI applied to green/sustainable supply chains, industry analysis and practice cases, and relevant industry policy documents.
Empirical grounding: illustrative industry cases and deployed tools cited (JD Logistics MRV‑T, Schneider Electric scheduling, Alibaba Cloud “Energy Expert”); no novel primary quantitative dataset or econometric estimation.
Methodological approach: pathway mapping organized by carbon management objectives (source reduction, operational efficiency, traceability, network collaboration); identification of structural limitations and prescriptive optimization directions combining technical and institutional solutions.

Implications for AI Economics

Externalities and accounting: economists and practitioners must internalize the full life‑cycle carbon costs of AI (training, inference, infrastructure) when assessing social net benefits—standard ex-ante and ex-post accounting frameworks are needed.
Market design and incentives: AI can reduce verification and monitoring costs, enabling more credible contracting (tiered procurement, performance‑based payments) and automated settlements (smart contracts). This alters principal–agent and multi‑party bargaining problems in supply chains and creates opportunities for mechanism design to align incentives across tiers.
Distributional effects and digital divide: firms with greater capital, data and AI capability (lead firms) may capture most benefits; policy or subsidy interventions may be justified to support SME digitalization and prevent a green digital divide that can worsen inequality in supply networks.
Credibility and adoption: explainability and interpretability are economic prerequisites for technology uptake—lack of trust imposes adoption frictions and slows diffusion; valuation models should include adoption costs tied to algorithmic transparency.
Rebound and counterfactuals: modeling efforts need to capture rebound effects where AI‑driven efficiency gains increase overall activity or are offset by AI’s own emissions. Welfare analyses should compare net emissions and economic welfare under different accounting boundaries and technology choices (e.g., lightweight vs. large models).
Policy and regulation implications: standardization of carbon data, interoperability rules, privacy‑preserving data sharing regulations, and certification/verification regimes for AI carbon performance will shape incentives and competition. Carbon markets and credits tied to verifiable AI‑enabled reductions may require new auditing protocols.
Research directions for applied economists:
- Quantify net emissions and cost‑benefit of AI interventions via empirical studies that include AI lifecycle emissions.
- Develop mechanism design and contract theory models for multi‑tier incentive alignment using AI‑enabled observability.
- Study adoption dynamics and welfare impacts of explainable vs. black‑box AI in managerial decision contexts.
- Model general equilibrium or network effects of large‑scale AI deployment in supply chains, including potential labor reallocation and investment complementarities.

Source: Ke, J. K. (2026). Pathways for AI‑Driven Green and Low‑Carbon Transition of Supply Chains. Modern Economy, 17(6):860–868. DOI: 10.4236/me.2026.176044.

Assessment

Paper Typetheoretical Evidence Strengthn/a — This is a conceptual/theoretical paper that proposes four AI-enabled pathways and policy/technical directions but provides no empirical tests, quantitative estimates, or causal identification of AI's effects on emissions. Methods Rigorn/a — No empirical methods are employed; the paper presents a conceptual framework and policy/technical recommendations based on synthesis rather than primary data or rigorous empirical evaluation. SampleNo empirical sample or dataset; the paper synthesizes existing literature and constructs a conceptual framework outlining four AI-driven pathways for supply‑chain decarbonization and associated structural limitations and policy/technical responses. Themesgovernance org_design innovation GeneralizabilityNot empirically validated — recommendations are theoretical and may not perform as claimed in practice, Applicability varies by industry, firm size, and supply‑chain complexity (e.g., retail vs. heavy manufacturing), Depends on digitalization and data-sharing readiness across supply‑chain partners and jurisdictions, Effectiveness contingent on local regulatory and institutional contexts and incentive structures, Quantitative trade‑offs (e.g., AI compute emissions vs. emission savings) are discussed qualitatively but not measured

Claims (15)

Claim	Direction	Outcome	Confidence & Evidence	Details
Global supply chains account for approximately 60% of total carbon emissions worldwide. Fiscal And Macroeconomic	null_result	share of total carbon emissions attributable to global supply chains	Reading fidelity high Study strength medium	not reported 0.12
Fragmented information across multiple actors and divergent interest objectives render systemic emission reductions unattainable through traditional management approaches. Governance And Regulation	negative	attainability of systemic emission reductions under traditional management	Reading fidelity high Study strength low	not reported 0.06
Intelligent demand sensing enhances supply–demand matching, curbing superfluous emissions at the source. Organizational Efficiency	positive	superfluous emissions at the source (via improved supply–demand matching)	Reading fidelity high Study strength speculative	not reported 0.02
Intelligent logistics scheduling optimizes the trade-off among cost, delivery time, and energy consumption. Organizational Efficiency	positive	trade-off among cost, delivery time, and energy consumption (including energy consumption reductions)	Reading fidelity high Study strength speculative	not reported 0.02
Chain-wide carbon footprint traceability addresses Scope 3 emissions. Regulatory Compliance	positive	Scope 3 emissions (supply-chain upstream/downstream emissions)	Reading fidelity high Study strength low	not reported 0.06
Supplier collaboration and empowerment promote green and low‑carbon transformation. Organizational Efficiency	positive	extent of green and low‑carbon transformation among suppliers	Reading fidelity high Study strength speculative	not reported 0.02
Structural limitations include fragmented data governance across organizations. Governance And Regulation	negative	degree of data governance fragmentation	Reading fidelity high Study strength low	not reported 0.06
The opacity of algorithmic decision‑making rationales is a structural limitation for AI-enabled carbon management. Ai Safety And Ethics	negative	opacity (lack of explainability) of algorithmic decision rationales	Reading fidelity high Study strength low	not reported 0.06
Organizational incentive misalignments hinder effective AI-driven carbon reduction across supply chains. Governance And Regulation	negative	degree to which organizational incentive misalignments impede carbon reduction	Reading fidelity high Study strength low	not reported 0.06
Environmental rebound risks arising from the computing power consumption of AI can offset emission reductions. Governance And Regulation	negative	net emissions after accounting for AI computing power consumption (environmental rebound effect)	Reading fidelity high Study strength low	not reported 0.06
Constructing a cross‑organizational carbon data governance framework is an effective optimization direction for AI‑driven supply‑chain decarbonization. Governance And Regulation	positive	effectiveness of cross‑organizational carbon data governance in enabling decarbonization	Reading fidelity high Study strength speculative	not reported 0.02
Developing explainable carbon decision intelligence models (i.e., explainable AI for carbon decisions) is a recommended optimization direction. Ai Safety And Ethics	positive	explainability of carbon decision models and resulting trust/use in governance	Reading fidelity high Study strength speculative	not reported 0.02
Designing collaborative carbon‑reduction incentive mechanisms that balance equity and efficiency will improve supply‑chain decarbonization. Governance And Regulation	positive	effectiveness of collaborative carbon‑reduction incentive mechanisms (balancing equity and efficiency)	Reading fidelity high Study strength speculative	not reported 0.02
Establishing a full‑life‑cycle carbon performance evaluation system is a recommended optimization direction for supply‑chain carbon governance. Regulatory Compliance	positive	coverage and quality of life‑cycle carbon performance evaluation	Reading fidelity high Study strength speculative	not reported 0.02
Embedding institutional design into technical processes enables AI to evolve from an efficiency tool into an institutional infrastructure for the green governance of supply chains. Governance And Regulation	positive	role of AI (efficiency tool vs. institutional infrastructure) in green governance of supply chains	Reading fidelity high Study strength speculative	not reported 0.02