U.S. time-series analysis links AI diffusion and renewable energy growth with long-run reductions in carbon intensity, while greater trade openness raises emissions intensity; however, results rest on aggregate national associations rather than clean causal identification.

From innovation to sustainability: Unravelling AI-driven solutions in the U.S.'s ecologically sustainable practices

Ratul Talukdar, Ahanaf Faiaz Mozumder, Tania Akter, Jewel Rana, Sabikun Nahar Khanshur, Md. Rashed, Md. Kamal Uddin · August 03, 2026 · Sustainable Futures

openalex correlational low evidence 7/10 relevance Summary only summary available; pdf_status=paywall 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

Ratul Talukdar provider ID
Ahanaf Faiaz Mozumder provider ID
Tania Akter provider ID
Jewel Rana provider ID
Sabikun Nahar Khanshur provider ID
Md. Rashed provider ID
Md. Kamal Uddin provider ID

Semantic Scholar

Latest observation: August 25, 2026

R. Talukdar provider ID
Ahanaf Faiaz Mozumder provider ID
Tania Akter provider ID
Jewel Rana provider ID
Sabikun Nahar Khanshur provider ID
Md. Rashed provider ID
Md. Kamal Uddin provider ID

Using U.S. annual data (1990–2022) and ARDL/VECM methods, the study reports that greater AI activity and higher renewable energy consumption are associated with lower carbon intensity in the long run, while trade openness increases carbon intensity and policy stringency and natural resource rents are statistically insignificant.

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

Since the United States (U.S.) is falling behind the net-zero carbon emissions target by 2050, understanding effective long-term policies to reduce carbon intensity/emissions is essential for the country. To address key literature gaps, this study aims to provide novel evidence of the comparative assessment of the long-term effectiveness of artificial intelligence (AI) patterns, environmental policy stringency (EPS), renewable energy consumption (REC), natural resource rents (NRR), and trade openness (TO) to facilitate the outcome of ecological sustainability (ECOI), which can either reduce or increase carbon emission intensity. This study is based on ecological modernization theory (EMT) and assumes these variables. The analysis employs the autoregressive distributed lag (ARDL) and the vector error correction model (VECM) based Granger causality test, which uses U.S. national-level data from 1990–2022 (33 years). The long-run ARDL findings suggest that AI and REC significantly enhance ECOI, lowering carbon emissions. Moreover, the NRR and EPS are insignificantly related to the result, but these two factors can also reduce emissions. Surprisingly, TO significantly reduces ECOI, increasing emissions in the long run, driven by carbon-intensive product imports and global trade-integrated industrial activities. These results are essential for federal policymakers to emphasize AI innovations and renewable transitions and control trade openness with eco-friendly policies on imports, enabling the country to reduce significant levels of carbon intensity/emissions in the long run and remain ahead of the net-zero-carbon target.

Summary

Main Finding

Using U.S. national data (1990–2022) and framed by ecological modernization theory, the study finds that artificial intelligence (AI) and renewable energy consumption (REC) significantly improve ecological sustainability (ECOI) in the long run (i.e., they reduce carbon-intensity/emissions). Natural resource rents (NRR) and environmental policy stringency (EPS) show statistically insignificant relationships with ECOI (though point estimates indicate they could reduce emissions). Trade openness (TO) has the opposite effect: it significantly worsens ECOI in the long run, increasing carbon intensity—attributed to carbon-intensive imports and trade-integrated industrial activity.

Key Points

AI adoption/patterns: Long-run beneficial effect on lowering carbon intensity; interpreted as AI enabling cleaner, more efficient production and energy use.
Renewable energy consumption (REC): Robust long-run emission-reducing effect consistent with decarbonization.
Environmental policy stringency (EPS): Insignificant in long-run estimates, but coefficients point toward emission reduction (possible delayed or indirect effects).
Natural resource rents (NRR): Insignificant but with a direction indicating potential to lower emissions (context-dependent).
Trade openness (TO): Statistically significant long-run increase in carbon intensity—likely due to imports of carbon-intensive goods and expansion of trade-linked emissions.
Causality: VECM-based Granger causality tests were used to assess directional relationships (study reports causality patterns though specific directions/details not given here).
Policy implication emphasis: Prioritize AI-driven innovations and renewable transition; manage trade openness with eco-friendly import and trade policies.

Data & Methods

Data: U.S. national-level annual data, 1990–2022 (33 observations).
Theoretical framework: Ecological Modernization Theory (EMT).
Econometric approach:
- Autoregressive Distributed Lag (ARDL) model to estimate short- and long-run relationships and cointegration among ECOI and regressors.
- Vector Error Correction Model (VECM)-based Granger causality tests to assess directionality and short-run versus long-run causation.
Variables: ECOI (ecological sustainability / carbon-intensity outcome), AI (measure of AI patterns/innovation), REC (renewable energy consumption), NRR (natural resource rents), EPS (environmental policy stringency), TO (trade openness). (Exact variable definitions and measurement units are study-specific.)
Notes on inference: Long-run ARDL estimates drive the headline conclusions; VECM used to support causality claims.

Implications for AI Economics

AI as a climate policy instrument: Evidence supports treating AI adoption and diffusion as a lever for long-term emissions reduction—AI investments can be justified not only by productivity gains but also by environmental benefits.
Targeting AI deployment: Policy should incentivize AI applications that yield measurable energy-efficiency and emissions-reduction outcomes (e.g., smart grids, demand-side management, process optimization in heavy industries).
Complementarity with renewables: AI and REC act synergistically—AI can enhance management, integration, and optimization of variable renewable resources, amplifying decarbonization returns.
Trade-policy interactions: Expansion of trade openness can offset domestic gains from AI/REC if imports are carbon-intensive. AI-focused climate policy should be coordinated with green trade measures (carbon-adjusted tariffs, border carbon adjustments, green procurement standards).
Regulatory design: Given EPS insignificance in these estimates, regulators should assess whether existing policy stringency has design or enforcement gaps; combining robust regulation with AI-enabled monitoring and compliance tools may improve policy effectiveness.
Research and measurement priorities: Future AI-economics work should improve measurement of "AI" (distinguish AI types, intensity, sectoral deployment), quantify magnitude of emission effects, and address potential endogeneity and sectoral heterogeneity to guide targeted interventions.

Assessment

Paper Typecorrelational Evidence Strengthlow — Findings rely on 33 annual national observations and time-series associations; ARDL/VECM and Granger tests provide information on long-run relationships and temporal ordering but do not establish causal mechanisms definitively—potential for omitted variable bias, reverse causation, measurement error in the AI variable, and sensitivity to specification or structural breaks. Methods Rigormedium — Use of ARDL and VECM is appropriate for nonstationary time series and cointegration analysis and is standard practice for separating short- and long-run effects, but rigor is limited by small sample size (33 observations), unclear variable construction (especially the AI measure), likely omitted confounders, and no strong strategy to address endogeneity beyond Granger causality. SampleU.S. national-level annual data covering 1990–2022 (33 observations); dependent variable is an ecological sustainability / carbon-intensity index (ECOI); regressors include an aggregate AI measure (AI adoption/innovation patterns), renewable energy consumption (REC), natural resource rents (NRR), environmental policy stringency (EPS), and trade openness (TO). Exact variable definitions and transformations were not provided in the supplied text. Themesadoption innovation governance IdentificationTime-series econometric approach using ARDL bounds testing to identify long-run cointegrating relationships and short-run dynamics, supplemented by VECM-based Granger causality tests to assess directional temporal relationships; no external instruments, natural experiment, or exogenous shock exploited for causal identification. GeneralizabilitySingle-country (United States) results may not hold for other countries with different energy mixes, industrial structures, or AI adoption patterns, Aggregate national-level analysis masks sectoral and regional heterogeneity (effects likely vary across industries and firm sizes), Annual frequency and 33 observations limit ability to detect short-term dynamics and may be sensitive to structural breaks or regime changes, Unclear measurement of the AI variable (type, intensity, sectoral deployment) reduces ability to generalize to specific AI applications, Findings reflect historical period (1990–2022); accelerating AI capabilities and policy changes after 2022 may alter relationships

Claims (6)

Claim	Direction	Outcome	Confidence & Evidence	Details
Artificial intelligence significantly improves ecological sustainability in the long run by reducing carbon intensity or emissions in the United States. Other	negative	Ecological sustainability (ECOI), operationalized as carbon intensity or emissions	Reading fidelity high Study strength medium	n=33 0.3
Renewable energy consumption significantly improves ecological sustainability in the long run by reducing emissions. Other	negative	Ecological sustainability (ECOI), measured through carbon intensity or emissions	Reading fidelity high Study strength medium	n=33 0.3
Environmental policy stringency has no statistically significant long-run relationship with ecological sustainability, although its point estimate indicates a possible reduction in emissions. Other	negative	Ecological sustainability (ECOI), measured through carbon intensity or emissions	Reading fidelity high Study strength low	n=33 0.15
Natural resource rents have no statistically significant long-run relationship with ecological sustainability, although the estimated direction suggests a potential reduction in emissions. Other	negative	Ecological sustainability (ECOI), measured through carbon intensity or emissions	Reading fidelity high Study strength low	n=33 0.15
Trade openness significantly worsens ecological sustainability in the long run by increasing carbon intensity. Other	positive	Carbon intensity or emissions as reflected in the ecological sustainability index (ECOI)	Reading fidelity high Study strength medium	n=33 0.3
The study uses VECM-based Granger causality tests to assess directional relationships among ecological sustainability, artificial intelligence, renewable energy consumption, natural resource rents, environmental policy stringency, and trade openness. Other	mixed	Directional and short-run versus long-run causality among the study variables	Reading fidelity high Study strength low	n=33 0.15