New quality productive forces driving green development under climate resilience regulation: an analysis of artificial intelligence enabled low-carbon energy transition

Low-carbon energy transition is an inevitable requirement of the concept of green development, reflecting the imperative to rationalize the metabolic exchange between humanity and nature. As a driver of new quality productive forces, artificial intelligence (AI) propels the green transformation of energy systems yet also raises concerns over “technological alienation”. This reveals its dual potential within specific production relations. However, systematic evidence remains lacking on whether climate resilience can moderate this tension. Using panel data from 269 prefecture-level cities in China (2008–2022), this study constructs a panel fixed-effects model that incorporates a quadratic term for AI development level and its interaction term with climate resilience. The moderating effect model is employed to examine how climate resilience influences the relationship between AI development and the low-carbon energy transition. Findings reveal that AI development initially inhibits but later facilitates energy decarbonization, while climate resilience significantly shifts the inflection point leftward, enabling highly resilient regions to bypass the initial suppression phase earlier. This illustrates how adaptive production relations can steer productive forces toward ecologically rational outcomes. Heterogeneity analysis further indicates this moderating effect is more pronounced in non-eastern regions, areas outside the Yangtze River Economic Belt, old industrial base regions and the Yellow River basin, aligning with the law of uneven development. By integrating asset lifecycle theory with digital technology, this study underscores climate resilience’s vital function in mitigating negative technological externalities and facilitating the rationalization of human-nature material metabolism. The findings provide theoretical and policy guidance for leveraging technology to empower green transformation and formulating regionally differentiated strategies to advance AI-driven decarbonization.

1 Introduction

Advancing the low-carbon energy transition of energy systems through large-scale integration of renewable energy and enhanced energy efficiency represents a critical pathway for addressing carbon emissions (Kong and Bai, 2023; Shen et al., 2024), as seen in key initiatives like the European Green Deal and the U.S. Infrastructure Investment and Jobs Act. Within this process, technology is recognized as one of the core drivers for achieving carbon neutrality (Zhao et al., 2025). General Secretary Xi Jinping has repeatedly emphasized the need to “deepen the application of digital technologies such as artificial intelligence (AI)” and “build a clean, low-carbon, safe, and efficient energy system.” This framing establishes a solid policy foundation for integrating digital intelligent technologies into the energy revolution. Artificial intelligence technology is viewed as a key tool for advancing the low-carbon energy transition of energy systems, yet its actual impact exhibits significant duality: On one hand, it can serve as a catalyst for the low-carbon energy transition by intelligently optimizing energy management and substantially enhancing system efficiency (Balaprakash and Dunn, 2021). On the other hand, AI itself can be carbon-intensive, especially during the initial deployment phase, due to high computational energy demands and reliance on carbon-heavy electricity grids (Kumari and Pandey, 2023). Effectively regulating the actual carbon impact of AI in energy transition to maximize its positive benefits has become a critical research topic in both international and regional contexts.

Despite its dual carbon impact, AI remains a key driver of the Fourth Industrial Revolution and holds great promise for the energy transition. AI is reshaping the global economic landscape and social structures with unprecedented depth and breadth (Lee et al., 2024). Through its powerful capabilities in data processing, pattern recognition, predictive analytics, and decision optimization, AI technology provides revolutionary tools for tackling complex challenges across industries (Rane et al., 2024). Particularly in the energy sector, AI is regarded as a key enabling technology to accelerate the low-carbon energy transition (Șerban and Lytras, 2020). Applications span optimizing generation forecasting and grid integration for intermittent renewables like wind and solar (Bhuiyan et al., 2025), enhancing energy efficiency in buildings, transportation, and industry (Lin and Yang, 2025; Zhou and Liu, 2024), to advancing the sophisticated management of smart grids, energy storage systems, and virtual power plants (Kong et al., 2024), and even accelerating the R&D process for new clean energy materials (Tabor et al., 2018), the application potential of AI permeates nearly every link in the energy value chain.

Marx’s historical materialism reminds us that technology is “by no means a simple factor of the productive forces, but rather the technical apparatus through which the relations of production reproduce themselves.” (Hoch et al., 1987) The ecological implications of AI hinge on its production dynamics, which are in turn influenced by material conditions. In reality, both AI itself and the energy infrastructure it depends on are deeply embedded in a physical environment that is increasingly affected by climate change. Consequently, this impact is constrained by the capacity of that environment to withstand climate shocks, that is, by its climate resilience. Furthermore, the entire lifecycle of AI, from research and development through training to large-scale deployment, relies heavily on a continuous, stable supply of energy and physical infrastructure (Alzoubi and Mishra, 2024). These systems, however, face severe challenges from climate change (Schwartzman, 2025). Climate resilience thus emerges not merely as an external factor but as a fundamental systemic attribute that directly regulates the extent to which AI can achieve carbon performance (Devarapalli, 2025). The IPCC emphasizes that measures such as investing in infrastructure and capacity are essential to adapt to projected climate impacts, making the building of climate-resilient societies a core task for ensuring sustainable development. Defined as the capacity to cope with climate variability (Barati et al., 2024), climate resilience may therefore serve as a critical mediating factor in the relationship between AI development and the low-carbon energy transition. A city with high climate resilience can provide foundational infrastructure support for low-carbon AI operations (Rane et al., 2024). Nevertheless, robust empirical evidence and theoretical frameworks are still lacking to demonstrate whether climate resilience systematically mediates AI’s impact on the transition to low-carbon energy.

Given this context, this paper employs fixed-effects and moderating models to examine the impact of climate resilience on the relationship between AI development and low-carbon energy transition using balanced panel data from China’s prefecture-level cities from 2008 to 2022. The findings reveal that AI exhibits a dynamic pattern of “initial suppression followed by promotion” in energy structure transformation. Climate resilience significantly shifts the inflection point of the original curve to the left, enabling highly resilient regions to bypass the initial suppression phase of the transition earlier. Heterogeneity analysis further indicates that this moderating effect is more pronounced in non-eastern regions, old industrial base regions and the Yellow River basin and areas outside the Yangtze River Economic Belt.

The marginal contribution of this paper lies in: First, it explains the key moderating role of climate resilience in the nonlinear mechanism through which AI development influences low-carbon energy transition, revealing the “context-dependence” of technology-enabled effects and expanding the research boundaries of climate governance and digital technology integration. Second, by modeling the impact of climate resilience on the relationship between AI development and low-carbon energy transition, it aids in understanding and addressing carbon footprints associated with large language model training and the AI development process itself (Pimenow et al., 2024). Furthermore, the identification of spatially heterogeneous moderating effects provides robust empirical evidence for formulating context-specific policies, thereby promoting a just and efficient nationwide low-carbon energy transition. Third, it refines the measurement framework for climate resilience, offering a reference tool for related research.

2 Literature review

2.1 Artificial intelligence development and low-carbon energy transition

AI, such as machine learning, deep learning, and computer vision, is being applied to the automated management and optimization of complex systems (Chen et al., 2021). The widespread adoption of AI is driving technological innovation, productivity gains, and socioeconomic development across industries (Ahmad et al., 2021; Zavyalova et al., 2023; Ding et al., 2024). In the energy sector, AI is recognized as a key enabling technology with the potential to optimize systems. Its application helps in managing intermittent renewable energy sources, enhancing grid stability, and improving energy efficiency (Nam et al., 2020; Višković et al., 2022; Lu et al., 2023; Saldanha et al., 2024; Wang et al., 2024b; Wang et al., 2025). However, AI itself is also an energy-intensive technology (Wang et al., 2024c). The development and operation of large-scale deep learning models consume substantial computational resources, directly translating into significant energy consumption and carbon emissions (Strubell et al., 2019). This creates a core contradiction: technologies designed to drive decarbonization may themselves become a major new source of energy demand, potentially challenging global efforts to reduce coal consumption and transition towards sustainable energy systems (Bogdanov et al., 2021; Shi et al., 2021).

This contradiction has led to divergent assessments of its net impact within academia, with empirical studies revealing a series of complex and nonlinear relationships. Some research identifies a “U-shaped” effect, suggesting that AI’s positive impacts may diminish at advanced stages (Zhao et al., 2024), while other studies support a “suppression-then-promotion” pathway, indicating that high carbon costs in the early stages may temporarily hinder the transition (Lee and Yan, 2024). Such nonlinear dynamics can be partially explained by mechanisms like the “energy rebound effect” (Xu et al., 2025). However, these explanations largely implicitly assume a stable external energy system and physical environment. This assumption overlooks how climate change increasingly threatens the reliability of the system’s foundations. Thus, a central theoretical and practical question emerges: Does the system’s “climate resilience”—its capacity to withstand climate shocks and ensure sustained operation—modulate the nonlinear relationship between AI development and low-carbon energy transition? Specifically, does it influence the critical inflection points and pathways from “inhibition” to “promotion”? Existing literature has yet to conduct systematic empirical tests of this moderating effect. Addressing this gap constitutes the core contribution of this study.

2.2 Climate resilience

Against the backdrop of global climate change, the frequency and intensity of climate-related disasters is increasing, with their impacts permeating multiple socioeconomic dimensions (Abbass et al., 2022). Climate change not only directly threatens urban safety and public health (Saeed et al., 2021) posing particularly severe challenges to rapidly urbanizing developing countries (Balk et al., 2009; Satterthwaite et al., 2012) but also exerts profound impacts on energy systems and economic development through a series of cascading effects.

The physical impacts of climate change directly threaten the infrastructure underpinning the energy transition. For instance, rising temperatures, extreme winds, sea-level rise, and reduced precipitation may affect the output and reliability of renewable energy sources (Osman et al., 2022). Such climate shocks frequently trigger power shortages, leading to multidimensional economic consequences: Power shortages not only reduce total factor productivity in enterprises (Guo et al., 2023) and increase corporate carbon emission intensity (Yu et al., 2023), but also negatively impact export profits (Bao et al., 2024). Developing renewable energy has proven to be a crucial pathway for mitigating energy insecurity and supporting economic growth (Xu et al., 2021). However, a key transition risk lies in the potential disruption of global efforts to limit coal consumption if new major energy-intensive sectors emerge (Shi et al., 2021), highlighting the dual challenge of maintaining energy system stability while steering toward low-carbon pathways.

In response to this systemic threat, building climate resilience—the capacity of social, economic, and environmental systems to withstand and adapt to climate disturbances while maintaining core functions—is recognized as a more effective strategic framework than implementing discrete adaptation measures (Tyler and Moench, 2012). Existing research on climate resilience has primarily focused on urban planning (Davoudi et al., 2012; Zhong and Li, 2023), disaster risk management (Hung et al., 2024), or the adaptability of specific infrastructure (Piemontese et al., 2024). However, its role in the critical process of low-carbon energy transition—particularly how it modulates the practical efficacy of digital technologies (such as AI) within this transition—remains under-explored (Lakhouit, 2025). This research gap implies that the theoretical role and empirical mechanisms of climate resilience as a potential key moderator variable linking climate risk, energy system stability, and technology-enabled low-carbon transition remain unclear.

2.3 Critical review

In summary, existing research has separately revealed the nonlinear characteristics of AI-driven low-carbon transformation and the threats posed by climate risks to energy system stability. However, these two strands have yet to be effectively connected. The vast majority of literature assesses AI’s impact by treating it as an abstract technology detached from specific material conditions. This approach fails to incorporate climate resilience—a core capability ensuring the sustained operation of systems—into the analytical framework as a key variable regulating the dynamic relationship between AI and the low-carbon energy transition. Consequently, systematically examining the moderating role of climate resilience in this relationship has become a central theoretical question demanding empirical validation. This study aims to fill this gap.

3 Mechanism of influence

3.1 The relationship between artificial intelligence development and low-carbon energy transition

The impact of AI development on low-carbon energy transition is not linear but stems from the dynamic equilibrium between its dual attributes as both an “energy-intensive technology” and an “enabling technology,” exhibiting a phased pattern of “initial suppression followed by subsequent promotion.” Technological factors can positively or negatively influence carbon emissions, ecological footprint, and energy transition (Wang et al., 2024a). In the early stages, AI’s high energy consumption and carbon emissions pose the primary contradiction: taking China as an example, under a coal-dominated energy structure, the surge in large-scale AI computing power directly translates into significant fossil fuel consumption, imposing a net burden on system decarbonization (Yu et al., 2021; Kumari and Pandey, 2023). However, as the technology matures, its enabling effects will surpass its own energy consumption to dominate the transformation process. This manifests primarily in two ways: First, direct energy savings and system optimization, such as using machine learning for precise regulation of building energy use or leveraging IoT to optimize urban energy flows (Balaprakash and Dunn, 2021; Goulart Tavares et al., 2021); Second, there should be comprehensive energy efficiency gains across industrial chains. The integration of AI and IoT significantly enhances efficiency in critical sectors like manufacturing and logistics (Fathi and Srinivasan, 2019; Gao et al., 2025). Their application extends further into scenarios such as wastewater treatment and ocean shipping, continuously unlocking energy-saving and carbon-reduction potential (Nam et al., 2020; Waltersmann et al., 2021; Abuella et al., 2023), fully demonstrating the vast prospects of intelligent systems driving deep decarbonization (Inderwildi et al., 2020).

Thus, the impact of AI development on the low-carbon energy transition is not linear but follows a pattern of initial suppression followed by promotion. This paper therefore proposes the following hypothesis 1: The impact of AI development on the low-carbon energy transition exhibits a dynamic pattern—initially negative, then positive.

3.2 The relationship between climate resilience and the effects of artificial intelligence development on low-carbon energy transition

Integrating Marxist political economy with asset lifecycle theory reveals that climate resilience, by transforming the material conditions of production, ensures the sustained realization of AI’s productive potential within the low-carbon transition. From the perspective of asset lifecycle theory, the core value of climate resilience lies in its ability to systematically enhance the positive contribution of AI assets to the low-carbon energy transition while reducing resistance throughout their entire lifecycle. This theory emphasizes that asset evaluation must cover the full cycle from planning, construction, and operation to maintenance and decommissioning (Treloar et al., 2000). It requires considering not only construction costs but also operation and maintenance risks. The “Limits to Growth” framework suggests that without climate resilience, the expansion of AI will face constraints. These include not only traditional resources like computing power and data but also energy and infrastructure bottlenecks intensified by climate shocks (Meadows et al., 1972). From an energy supply perspective, climate shocks such as extreme weather increasingly threaten energy availability. This amplifies the high-carbon footprint of AI development (Gonçalves et al., 2024). Power shortages triggered by extreme weather are becoming more frequent. Many regions globally, including coal-dependent China, often rely on high-carbon power units to meet peak demand during shortages (Zhao et al., 2025). This reinforces carbon lock-in within power systems and delays the shift to cleaner alternatives. Regarding infrastructure, climate change directly threatens physical security and triggers cascading functional failures. Shocks like floods submerging data centers or heatwaves disrupting cooling systems can cause operational halts and damage to hardware (Pasupuleti, 2025). This disruption halts the green services enabled by AI, resulting in severe “transition opportunity losses.” These losses refer to the unrealized transformative benefits—such as energy efficiency gains and renewable integration—that AI optimization could have delivered, thereby objectively slowing the overall decarbonization of the system.

However, embedding climate resilience from the outset of an asset’s lifecycle changes this dynamic. Proactive measures include selecting sites to avoid high-risk areas, adopting higher-standard protective designs, and building flexible distributed energy systems. These significantly enhance an asset’s stable operation under climate stress. Investing during planning and construction addresses the climate stresses the asset will face over decades of operation. The core benefit is a substantial reduction in the frequency and severity of climate-induced operational disruptions. This safeguards the continuity and reliability of AI-driven low-carbon services. The advantages manifest as significantly fewer climate-related operational interruptions, less high-carbon regression, and minimized loss of service functionality over the asset’s full lifecycle. This reduces recurring high-carbon setbacks caused by climate impacts. By ensuring the continuous, stable delivery of AI-powered low-carbon services, climate resilience minimizes unnecessary disruptions to the energy system’s transition. It significantly increases the net positive benefits of AI as a transformational enabler and accelerates its contribution to deep decarbonization.

Based on the logic that “front-end resilience investments reduce back-end risk costs,” this paper proposes the following hypothesis 2: Higher climate resilience shortens the net suppression phase of AI and accelerates the emergence of its transformative benefits at earlier developmental stages.

4 Model construction and variable description

4.1 Model construction

This study first examines whether AI development has a nonlinear relationship with the low-carbon energy transition, and the model is constructed as shown in Equation 1:

${transition}_{it}={\alpha }_0+{\alpha }_1{AI}_{it}+{\alpha }_2{AI}_{it}^2+\sum {\theta }_j{Control}_{jit}+{\mu }_i+{\nu }_t+{\epsilon }_{it}(1)$

Among these, transition_it denotes the low-carbon energy transition, AI_it represents AI development, α₂ is the core estimated coefficient, and Control_jit constitutes the series of control variables. Additionally, α₀ denotes the constant term, μ_i and ν_t represent city-specific and time-specific fixed effects respectively, while ε_it constitutes the random error term.

Furthermore, this paper constructs a panel fixed-effects model that incorporates a quadratic term for AI development level and its interaction term with climate resilience, as shown in Equation 2. It examines whether climate resilience influences the relationship between AI development and the low-carbon energy transition by constructing the following model:

$\begin{array}{c}{transition}_{it}={\alpha }_0+{\alpha }_1{AI}_{it}+{\alpha }_2{AI}_{it}^2+{\alpha }_3{Climaresil}_{it}+{\alpha }_4{Climaresil}_{it}\times {AI}_{it}\\ +{{\alpha }_{5}Climaresil}_{it}\times {AI}_{it}^2+\sum {\theta }_j{Control}_{jit}+{\mu }_i+{\nu }_t+{\epsilon }_{it}\end{array}(2)$

Among these, transition_it denotes the low-carbon energy transition, AI_it represents AI development, Climaresil_it signifies climate resilience, and α₅ is the core parameter to be estimated. Control_jit denotes the series of control variables. Additionally, α₀ denotes the constant term, μ_i and ν_t represent city-specific and time-specific fixed effects respectively, while ε_it constitutes the random error term.

4.2 Variable declaration

4.2.1 Dependent variable

Low-Carbon Energy Transition (transition). Low-carbon energy transition refers to the continuous process of optimizing and adjusting dominant energy sources toward low-carbon alternatives through substitution and complementarity. As China accelerates the clean transformation of energy consumption, its energy structure exhibits an adjustment trend characterized by replacing high-carbon sources with green, low-carbon alternatives. Therefore, relying solely on the proportion of coal or clean energy consumption to measure the progress of low-carbon energy transition is inadequate. Therefore, this paper adopts the methodology proposed by Fu (2010) to construct an index measuring the low-carbon energy transition. Considering data availability, this paper categorizes energy sources into coal, oil and gas, and other energy sources, and their respective shares are treated as components of a spatial vector, forming a set of three-dimensional vectors: $E_t=\left({\mathit{e}}_t^1,{\mathit{e}}_t^2,{\mathit{e}}_t^3\right)$.

Then, as shown in Equation 3, calculate the angles ${\theta }_t^1$, ${\theta }_t^2$, ${\theta }_t^3$ between $E_t$ and the vectors of energy consumption arranged from high carbon to low carbon $E_0^1=\left(1,0,0\right)$, $E_0^2=\left(0,1,0\right)$, $E_0^3=\left(0,0,1\right)$:

${\theta }_t^n=\arccos \left(\frac{{\displaystyle {\sum }_{i=1}^3}\left({\mathit{e}}_t^i\times {\mathit{e}}_0^i\right)}{{\left({{\displaystyle {\sum }_{i=1}^3}\left({\mathit{e}}_t^i\right)}^2\times {{\displaystyle {\sum }_{i=1}^3}\left({\mathit{e}}_0^i\right)}^2\right)}^{1/2}}\right)n=1,2,3(3)$

Finally, as shown in Equation 4, the angles between all vectors are weighted to form the low-carbon energy transition Index $\text{transition}$, with the specific calculation formula shown below:

$transition={\displaystyle \sum _{s=1}^3}{\displaystyle \sum _{t=1}^s}{\theta }_t^n(4)$

4.2.2 Independent variable

Artificial Intelligence Development (AI). This paper adopts the methodology of Wei et al. (2020), using the logarithm of the stock of AI enterprises as an indicator of AI development.

4.2.3 Moderating variable

Climate Resilience (Climaresil). Initially defined as “the capacity of societies, economies, and ecosystems to respond to climate events, trends, or disturbances,” climate resilience has evolved through research from a singular concept of disaster resistance into a comprehensive framework encompassing system stability, adaptability, and transformative capacity. This study, through reviewing relevant literature (Tyler et al., 2016; Summers et al., 2017; Li and Wang, 2023; Wang and Chen, 2024), finds that most climate resilience indicator systems are extensions of urban resilience frameworks applied to climate contexts. This approach lacks temporal dynamics and process orientation. Aligning with the definition of climate resilience, this study constructs a framework organized chronologically—pre-disaster, during disaster, and post-disasterv—to capture capabilities across different stages of climate disaster response. The framework comprises four primary indicators: “Predictive Capacity,” “Adaptation and Response Capacity,” “Stability and Recovery Capacity,” and “Learning and Transformation Capacity.” Entropy analysis is employed to measure climate resilience across prefecture-level cities, with specific secondary indicators are detailed in Table 1.

Table 1

Table 1. Municipal climate resilience index system.

4.2.4 Control variables

This study follows the methodology of previous research (Shahbaz et al., 2022; Feng et al., 2024; Xu et al., 2025) by selecting the following control variables: Industrial Structure Upgrading, Fiscal Pressure, Financial Development, Highway Freight Volume, Sulfur Dioxide Concentration, and Urbanization.

4.3 Data sources

This study examines 269 municipal-level administrative regions from 2008 to 2022, with data primarily sourced from four categories. Low-carbon energy transition data originate from the China Energy Statistical Yearbook, provincial statistical yearbooks, and municipal statistical yearbooks. AI development data are derived from China FIR robotics data and the China AI Enterprise Database. Climate resilience data includes: Fixed asset investment in municipal public infrastructure construction (10,000 yuan), local general public budget expenditure (100 million yuan), number of higher education students per 10,000 people, and general public budget expenditure on science and technology as a percentage of GDP (%) are sourced from the National Statistical Yearbook, provincial statistical yearbooks, municipal statistical yearbooks, and the Urban Construction Statistical Yearbook. Comprehensive television program population coverage (%) is sourced from provincial statistical yearbooks, economic statistical yearbooks, and municipal statistical bulletins on national economic and social development. Green patent applications (units) are sourced from the China National Research Data Service (CNRDS). Industrial structure rationalization is calculated using data from provincial statistical yearbooks and municipal statistical yearbooks. Control variables are sourced from provincial statistical yearbooks and municipal statistical yearbooks. Cities with excessive missing values were excluded from this study. The data can be found in the Supplementary Data Sheet 1. Descriptive statistics for each variable are presented in Table 2.

Table 2

Table 2. Descriptive statistics analysis of variables.

5 Results

5.1 Benchmark regression analysis

Table 3 presents the results of the benchmark regression in columns (1) to (4). In terms of model specification, columns (1) and (2) do not include time or individual fixed effects, while columns (3) and (4) incorporate both. The preliminary results in columns (1) to (3) show that the squared term of AI development (AI²) has a positive coefficient for low-carbon energy transition (transition), which is significant at the 10% level. After introducing additional control variables to reduce omitted variable bias, the coefficient of the squared term of AI development (AI²) on low-carbon energy transition (transition) is 0.006 and significant at the 5% level. This suggests a potential nonlinear relationship between AI development and the low-carbon energy transition. Following the three-step test by Lind and Mehlum, the U-test identifies an inflection point at 4.959 within the observed range of AI development. The slope of the curve is negative to the left of this point and positive to the right. The slopes on both sides are significant and opposite in direction, confirming the nonlinear pattern. The overall test is significant at the 5% level. The presence of this inflection point indicates that when AI development is below 4.959, it inhibits the low-carbon energy transition. Once AI development exceeds this threshold, it exerts a positive driving effect. During the sample period, approximately 86% of cities completed at least one transition from left to right, indicating that the AI development in most Chinese cities has crossed a critical inflection point. Therefore, hypothesis one is supported: the impact of AI development on the low-carbon energy transition follows a dynamic pattern—negative initially, then positive.

Table 3

Table 3. Benchmark regression results.

5.2 Moderating effect analysis

This study examines the moderating role of climate resilience in the relationship between AI development and the low-carbon energy transition. The results are presented in column (1) of Table 4. The interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) is significantly positive. Following the criterion proposed by Haans et al. (2016), the calculation ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.001028<0$ confirms that an increase in the moderating variable shifts the inflection point to the left. This demonstrates a significant moderating effect of climate resilience, thereby validating hypothesis 2: higher climate resilience shortens the initial net inhibitory phase of AI and accelerates the emergence of its positive effects on the low-carbon energy transition at earlier stages of development.

Table 4

Table 4. Moderating effect analysis.

5.3 Robustness tests

This paper employs endogeneity tests, substitution of the dependent variable, sample truncation at the 5% tail, and exclusion of municipalities directly under the central government to validate the robustness of the conclusions.

5.3.1 Endogeneity test

It should be noted that the model in this study may be subject to endogeneity concerns. To further address this issue, we employ an instrumental variable (IV) approach. Following the method of Chen et al. (2025), which uses historical variables as instruments, we construct an IV based on the 2006 urban built-up area drainage pipe density for each city, interacted with an indicator for whether the city was selected for a pilot policy in 2024. The selection of this IV is justified for two main reasons. First, it satisfies the relevance condition. Climate resilience development demonstrates strong path dependence and state dependence. The 2006 drainage pipe density and the 2024 pilot policy for deepening climate-resilient city development are inherently correlated with subsequent levels of climate resilience, ensuring instrument relevance. Second, the exogeneity condition is also met. The 2006 drainage pipe density is unlikely to have a direct effect on the low-carbon energy transition after 2008. Similarly, the 2024 pilot policy could not have influenced the transition before 2022. As these historical variables are time-invariant, we interact with the lagged climate resilience variable to form the instrument for regression estimation.

Table 5 reports the Two-Stage Least Squares (2SLS) regression results. The first-stage results confirm a significant relationship between the instrumental variables and the core explanatory variable. The KP-F statistic is 10.19, which exceeds the critical value of 8.96 at a 15% bias level, supporting the strength of the instruments and satisfying the relevance condition. In the second stage, after addressing endogeneity, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) remains significantly positive. With ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.02464<0$, an increase in the moderating variable the inflection point continues to shift to the left.

Table 5

Table 5. Endogeneity test.

5.3.2 Substitution of the dependent variable

The dependent variable “low-carbon energy transition” was replaced with “coal share.” If the regression result satisfies ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4<0$, it would indicate that climate resilience significantly shifts the inflection point of the original curve to the left. The regression results are presented in column (2) of Table 4. The table shows that the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) is significantly negative, with ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.000066<0$. This confirms that enhanced climate resilience shifts the inflection point leftward. The finding implies that in regions with higher climate resilience, AI development reaches the turning point earlier. This accelerates the decline in the share of coal consumption and thus speeds up the decarbonization of the energy structure.

5.3.3 Sample truncation at the 5% tail

To reduce the potential influence of outliers on the regression results, the dependent variable was winsorized at the fifth percentile. The corresponding regression results are shown in column (3) of Table 4. As indicated in the table, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) remains significantly positive. The expression ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.000503<0$ confirms that increasing the moderator variable shifts the inflection point to the left. This result demonstrates that after mitigating the distortion caused by outliers, the main conclusions of the study continue to hold.

5.3.4 Exclude municipalities directly under the central government

Municipalities directly under the central government differ considerably from other prefecture-level cities in economic structure, political influence, and resource allocation. These distinct features may cause them to appear as outliers or extreme values in research variables. To address this, municipalities are excluded from the sample, and the regression is re-estimated. The results are reported in column (4) of Table 4. As shown, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) remains significantly positive. The calculation ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.001944<0$ indicates that raising the moderator variable shifts the inflection point to the left. This confirms that, even after reducing the potential distortion caused by municipalities, the core findings of this paper remain valid.

5.4 Heterogeneity analysis

The Marxist theory of uneven development posits that the advancement of productive forces and production relations often exhibits systemic disparities characterized by asynchrony and heterogeneity across different regions. China’s regions exhibit significant disparities in economic development, technological advancement, and policy frameworks. As a result, the moderating effect of climate resilience on the AI and low-carbon energy transition relationship may also vary spatially. Overlooking this heterogeneity could lead to a distorted assessment of climate resilience’s role and potentially worsn regional development imbalances. A city’s geographical location serves as a composite indicator, reflecting the combined influence of economic, technological, and policy factors. Therefore, this study adopts membership in the Yangtze River Economic Belt, urban location, old industrial base regions and Yellow River basinas the analytical framework to examine this heterogeneous impact. The sample is grouped accordingly for heterogeneity analysis, and the regression results are presented in Tables 6,7.

Table 6

Table 6. Heterogeneity analysis results.

Table 7

Table 7. Heterogeneity analysis results.

5.4.1 Yangtze river economic belt

The impact of climate resilience on the relationship between AI and the low-carbon energy transition may also exhibit spatial heterogeneity. To examine regional variations, this study divides the sample areas into the Yangtze River Economic Belt and other regions for further testing. The regression results are shown in Table 6. After controlling for two fixed effects, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) is significantly positive in regions outside the Yangtze River Economic Belt. The expression ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.015485<0$ indicates that climate resilience shifts the inflection point to the left. In contrast, the corresponding interaction term (Climaresil × AI²) for the Yangtze River Economic Belt is not statistically significant. One possible explanation is that the Yangtze River Economic Belt, as a core economic region in China, is more economically developed and industrialized, and may have already entered a later stage of low-carbon energy transition. AI applications in this region could be relatively mature, reducing their marginal impact on the transition and thus weakening the observable moderating effect of climate resilience. In regions outside the Yangtze River Economic Belt, economic development is generally less advanced, and the low-carbon energy transition is likely still in its early or middle stages. Here, AI development of is more sensitive to the progress of the transition. Climate resilience can shift the original inflection point leftward because it helps these regions better withstand external shocks during the transition, thereby amplifying the effects of AI.

5.4.2 Eastern region

This paper divides the sample regions into eastern and non-eastern areas for further heterogeneity analysis. The regression results are presented in Table 6. After controlling for two fixed effects, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) in non-eastern regions is significantly positive. The calculation ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.010951<0$ indicates that climate resilience shifts the inflection point to the left. In contrast, the corresponding interaction term (Climaresil × AI²) in eastern regions is not statistically significant. This may be because energy systems in eastern regions, though large and complex, benefit from relatively well-developed infrastructure. Consequently, AI-driven optimization of these energy systems can operate with relative stability even without additional climate resilience measures, making the moderating effect of climate resilience less pronounced. Non-eastern regions, however, possess weaker energy infrastructure and more vulnerable systems. This makes them more susceptible to disruptions from climate risks such as extreme weather. In such contexts, improving climate resilience can significantly enhance the potential of AI technologies.

5.4.3 Old industrial base regions

This study divides the sample area into old industrial base regions and non-old industrial base regions for heterogeneity analysis. The regression results are presented in Table 7. After controlling for two fixed effects, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) in old industrial base regions is significantly positive. The calculation ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.025236<0$ indicates that climate resilience shifts the inflection point to the left. In contrast, the corresponding interaction term (Climaresil × AI²) in non-old industrial base regions is not statistically significant. This may stem from older industrial bases being in a “green transition catch-up phase,” where the combination of AI and climate resilience can more rapidly overcome technological or institutional bottlenecks, yielding significant synergistic effects. Non-older industrial bases may have entered a stable development phase, where the interaction between the two factors has reached saturation or has yet to achieve economies of scale.

5.4.4 Yellow river basin

This study divides the sample area into cities within the Yellow River and those outside it for heterogeneity analysis. The regression results are presented in Table 7. After controlling for two fixed effects, the interaction term between the quadratic term of AI and climate resilience (Climaresil × AI²) in cities within the Yellow River is significantly positive. The calculation ${\mathrm{\alpha }}_1\times {\mathrm{\alpha }}_5-{\mathrm{\alpha }}_2\times {\mathrm{\alpha }}_4=-0.017499<0$ indicates that climate resilience shifts the inflection point to the left. In contrast, the corresponding interaction term (Climaresil × AI²) in other cities is not statistically significant. This may stem from the complex energy systems in Yellow River cities, which are highly vulnerable to climate change impacts. Consequently, additional climate resilience measures are required to ensure relatively stable operation of AI-driven energy system optimization. Against this backdrop, enhancing climate resilience significantly amplifies the application potential of AI technologies.

6 Policy implications

This study empirically examines the moderating role of climate resilience in the relationship between AI development and low-carbon energy transition. It uses panel data from 269 prefecture-level cities in China from 2008 to 2022. Key findings are as follows: The impact of AI development on low-carbon energy transition shows a significant nonlinear trend. This trend is characterized by “initial suppression followed by promotion.” Climate resilience is a core capability for addressing climate risks. It significantly shifts the inflection point of this curve to the left. Heterogeneity analysis reveals significant spatial variation in this moderating effect.

This study reveals that climate resilience, as an embodiment of adaptive production relations, can effectively guide AI technology—a new form of productive force—toward ecological rationality. Based on these findings, the paper proposes the following policy recommendations: First, according to the key findings, the effect of AI on the low-carbon energy transition follows a dynamic pattern, initially negative, then positive. Therefore, evaluations must move beyond static perspectives. Governments should establish long-term, dynamic observation and assessment frameworks. Strategic patience is essential in the early stages. Even if the energy consumption structure faces temporary pressure, continued investment in AI R&D and infrastructure remains crucial. This investment will unlock AI’s full long-term emission reduction potential.

Second, national and local governments should explicitly make enhanced climate resilience a core objective. According to the key findings, climate resilience significantly shifts the curve’s inflection point leftward, demonstrating its central role in mitigating climate risks. This should guide AI industry development plans and low-carbon energy transition roadmaps. We recommend establishing a comprehensive climate resilience assessment system. This system should cover the entire chain of “prediction-response-recovery-learning.” It should then serve as a key metric for evaluating AI projects, smart city development, and new energy systems. This approach will help prevent climate risks from disrupting the low-carbon energy transition at the top-level design stage.

Third, policies for climate resilience must account for regional differences. Aoide a one-size-fits-all approach. According to the key findings, climate resilience only exhibits a significant leftward shift in the relationship between AI and low-carbon energy transition outside the Yangtze River Economic Belt, eastern regions, old industrial base regions and Yellow River basin, while its moderating effect is negligible in relatively developed areas. The most appropriate local measures should be selected. Regions like the Yangtze River Economic Belt and eastern China have strong digital foundations. Their transformation is approaching an inflection point. Efforts there should focus on deepening the integration of AI in the energy sector. This will accelerate their regional energy systems past critical transition thresholds. Other regions have weaker resilience foundations. Their policy priorities should focus on boosting local energy infrastructure resilience, digitalization rates, and human capital levels. These regions should introduce AI cautiously and in an orderly manner. This guards against the risks of energy consumption lock-in or a shift toward higher carbon intensity. For old industrial base regions, barracks should drive green transformation through “intelligent reconstruction” to avoid lock-in risks. For the Yellow River basin region, AI empowerment should be constrained by “ecological priority” to strengthen systemic coordination.

Author contributions

BS: Conceptualization, Writing – original draft, Supervision, Writing – review and editing. ZW: Methodology, Writing – review and editing, Writing – original draft, Data curation, Formal Analysis. GL: Conceptualization, Writing – review and editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the National Natural Science Foundation of China (Grants No. 72374063 and 72304089), the Natural Science Foundation of Hebei Province (Grant No. G2025207010), the Humanities and Social Sciences Research Project for Universities in Hebei Province (Grant No. BJS2024068), the Science Research Project of the Hebei Education Department (Grant No. QN2026325), and the Social Science Development Research Project of Hebei Province (Grant No. 20230303002).

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1765675/full#supplementary-material

References

Abbass, K., Qasim, M. Z., Song, H., Murshed, M., Mahmood, H., and Younis, I. (2022). A review of the global climate change impacts, adaptation, and sustainable mitigation measures Environ. Sci. Pollut. Res. Int. 29 (28), 42539–42559. doi:10.1007/s11356-022-19718-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Abuella, M., Atoui, M. A., Nowaczyk, S., Johansson, S., and Faghani, E. (2023). “Data-driven explainable artificial intelligence for energy efficiency in short-sea shipping,” in Joint european conference on machine learning and knowledge discovery in databases (Cham: Springer Nature Switzerland), 226–241. doi:10.1007/978-3-031-43430-3_14

CrossRef Full Text | Google Scholar

Ahmad, T., Zhang, D., Huang, C., Zhang, H., Dai, N., Song, Y., et al. (2021). Artificial intelligence in sustainable energy industry: Status quo, challenges and opportunities. J. Cleaner Production 289, 125834. doi:10.1016/j.jclepro.2021.125834

CrossRef Full Text | Google Scholar

Alzoubi, Y., and Mishra, A. (2024). Green artificial intelligence initiatives: potentials and challenges. J. Clean. Prod. 468, 143090. doi:10.1016/j.jclepro.2024.143090

CrossRef Full Text | Google Scholar

Balaprakash, P., and Dunn, J. B. (2021). “Overview of data science and sustainability analysis,” in Data science applied to sustainability analysis (Elsevier), 1–14. doi:10.1016/B978-0-12-817976-5.00001-2

CrossRef Full Text | Google Scholar

Balk, D., Montgomery, M. R., McGranahan, G., Kim, D., Mara, V., Todd, M., et al. (2009). Mapping urban settlements and the risks of climate change in Africa, Asia and South America. Popul. Dyn. Clim. Change 80, 103. Available online at: https://www.iied.org/g02650.

Google Scholar

Bao, B., Fu, D., Yu, J., and Zhang, Y. (2024). Lights dim, exports down: examining the trade effects of power shortages on Chinese manufacturing firms. China Econ. Rev. 88, 102270. doi:10.1016/j.chieco.2024.102270

CrossRef Full Text | Google Scholar

Barati, M., Soundharajan, B., and Nikoo, M. (2024). Simulation of climate-adaptation responses to rainfall variability on rainfed yield anomalies. Environ. Sustain. Indic. 22, 100411. doi:10.1016/j.indic.2024.100411

CrossRef Full Text | Google Scholar

Bhuiyan, S., Chowdhury, A., Hossain, M., Mobin, S., and Parvez, I. (2025). AI-driven optimization in renewable hydrogen production: a review. Am. J. Interdiscip. Stud. 6 (1), 76–94. doi:10.63125/06z40b13

CrossRef Full Text | Google Scholar

Bogdanov, D., Ram, M., Aghahosseini, A., Gulagi, A., Oyewo, A. S., Child, M., et al. (2021). Low-cost renewable electricity as the key driver of the global energy transition towards sustainability. Energy 227, 120467. doi:10.1016/j.energy.2021.120467

CrossRef Full Text | Google Scholar

Chen, C., Hu, Y., Karuppiah, M., and Kumar, P. M. (2021). Artificial intelligence on economic evaluation of energy efficiency and renewable energy technologies. Sustain. Energy Technol. Assessments 47, 101358. doi:10.1016/j.seta.2021.101358

CrossRef Full Text | Google Scholar

Chen, X., Qiu, B., and Ji, Z. (2025). Urban network embeddedness, information availability, and firm innovation boundary. Econ. Res. J. 60 (6), 57–76. doi:10.3969/j.issn.0577-9154.2025.06.005

CrossRef Full Text | Google Scholar

Davoudi, S., Shaw, K., Haider, L. J., Quinlan, A. E., Peterson, G. D., Wilkinson, C., et al. (2012). Resilience: a bridging concept or a dead end?“Reframing” resilience: challenges for planning theory and practice interacting traps: resilience assessment of a pasture management system in Northern Afghanistan urban resilience: what does it mean in planning practice? Resilience as a useful concept for climate change adaptation? The politics of resilience for planning: a cautionary note. Plan. Theory and Pract. 13 (2), 299–333. doi:10.1080/14649357.2012.677124

CrossRef Full Text | Google Scholar

Devarapalli, P. (2025). Patent or planet? Rethinking IP for just climate tech transition in the global south. Glob. Environ. Polit. 25 (4), 128–150. doi:10.1162/glep.a.25

CrossRef Full Text | Google Scholar

Ding, T., Li, H., Liu, L., and Feng, K. (2024). An inquiry into the nexus between artificial intelligence and energy poverty in the light of global evidence. Energy Econ. 136, 107748. doi:10.1016/j.eneco.2024.107748

CrossRef Full Text | Google Scholar

Fathi, S., and Srinivasan, R. (2019). “Climate change impacts on campus buildings energy use: an AI-based scenario analysis,” in Proceedings of the 1st ACM international workshop on urban building energy sensing, controls, big data analysis, and visualization, 112–119. doi:10.1145/3363459.336354

CrossRef Full Text | Google Scholar

Feng, C., Ye, X., Li, J., and Yang, J. (2024). How does artificial intelligence affect the transformation of China’s green economic growth? An analysis from internal-structure perspective. J. Environ. Manag. 351, 119923. doi:10.1016/j.jenvman.2023.119923

PubMed Abstract | CrossRef Full Text | Google Scholar

Fu, L. (2010). An empirical research on industry structure and economic growth. Stat. Res. 27 (8), 79–81. doi:10.19343/j.cnki.11-1302/c.2010.08.011

CrossRef Full Text | Google Scholar

Gao, X., Ji, X., Wang, R., and Yu, J. (2025). The effect of artificial intelligence on energy transition: evidence from China. Energy Econ. 147, 108568. doi:10.1016/j.eneco.2025.108568

CrossRef Full Text | Google Scholar

Gonçalves, A. C., Costoya, X., Nieto, R., and Liberato, M. L. (2024). Extreme weather events on energy systems: a comprehensive review on impacts, mitigation, and adaptation measures. Sustain. Energy Res. 11 (1), 4. doi:10.1186/s40807-023-00097-6

CrossRef Full Text | Google Scholar

Goulart Tavares, G., Capriles, P. V., and Goliatt, L. (2021). “Automatic evolutionary settings of machine learning methods for buildings’ thermal loads prediction,” in EPIA conference on artificial intelligence (Cham: Springer International Publishing), 183–195. doi:10.1007/978-3-030-86230-5_15

CrossRef Full Text | Google Scholar

Guo, D., Li, Q., Liu, P., Shi, X., and Yu, J. (2023). Power shortage and firm performance: evidence from a Chinese city power shortage index. Energy Econ. 119, 106593. doi:10.1016/j.eneco.2023.106593

CrossRef Full Text | Google Scholar

Haans, R. F., Pieters, C., and He, Z. L. (2016). Thinking about U: theorizing and testing u-and inverted u-shaped relationships in strategy research. Strategic Manag. J. 37 (7), 1177–1195. doi:10.1002/smj.2399

CrossRef Full Text | Google Scholar

Hoch, P., Mackenzie, D., and Wajcman, J. (1987). The social shaping of technology. Technol. Cult. 28 (1), 132. doi:10.2307/3105489

CrossRef Full Text | Google Scholar

Hung, C., Hung, H., and Hsu, M. (2024). Linking the interplay of resilience, vulnerability, and adaptation to long-term changes in metropolitan spaces for climate-related disaster risk management. Clim. Risk Manag. 44, 100618. doi:10.1016/j.crm.2024.100618

CrossRef Full Text | Google Scholar

Inderwildi, O., Zhang, C., Wang, X., and Kraft, M. (2020). The impact of intelligent cyber-physical systems on the decarbonization of energy. Energy and Environ. Sci. 13 (3), 744–771. doi:10.1039/C9EE01919G

CrossRef Full Text | Google Scholar

Kong, Z., and Bai, K. (2023). “Optimizing carbon emission reduction strategies through AI-enabled modeling: paving the way to a sustainable future,”, 12793. SPIE, 178–184. doi:10.1117/12.3006441Int. Conf. Mechatronics Intelligent Control (ICMIC 2023)

CrossRef Full Text | Google Scholar

Kong, J., Dong, Y., Zhang, Z., Yap, P. S., and Zhou, Y. (2024). Advances in smart cities with system integration and energy digitalization technologies: a state-of-the-art review. Sustain. Energy Technol. Assessments 72, 104012. doi:10.1016/j.seta.2024.104012

CrossRef Full Text | Google Scholar

Kumari, N., and Pandey, S. (2023). “Application of artificial intelligence in environmental sustainability and climate change,” in Visualization techniques for climate change with machine learning and artificial intelligence (Elsevier), 293–316. doi:10.1016/B978-0-323-99714-0.00018-2

CrossRef Full Text | Google Scholar

Lakhouit, A. (2025). Investigating the hydrogen renaissance in the global energy transition with AI integration. Energy Convers. Manag. 26, 101010. doi:10.1016/j.ecmx.2025.101010

CrossRef Full Text | Google Scholar

Lee, C. C., and Yan, J. (2024). Will artificial intelligence make energy cleaner? Evidence of nonlinearity. Appl. Energy 363, 123081. doi:10.1016/j.apenergy.2024.123081

CrossRef Full Text | Google Scholar

Lee, C. C., Fang, Y., Quan, S., and Li, X. (2024). Leveraging the power of artificial intelligence toward the energy transition: the key role of the digital economy. Energy Econ. 135, 107654. doi:10.1016/j.eneco.2024.107654

CrossRef Full Text | Google Scholar

Li, G., and Wang, L. (2023). Study of regional variations and convergence in ecological resilience of Chinese cities. Ecol. Indic. 154, 110667. doi:10.1016/j.ecolind.2023.110667

CrossRef Full Text | Google Scholar

Lin, B. Q., and Yang, Y. X. (2025). Building efficiency: how the national AI innovation pilot zones enhance green energy utilization? Evidence from China. J. Environ. Manag. 387, 125945. doi:10.1016/j.jenvman.2025.125945

PubMed Abstract | CrossRef Full Text | Google Scholar

Lu, L., Liu, P., Yu, J., and Shi, X. (2023). Digital inclusive finance and energy transition towards carbon neutrality: evidence from Chinese firms. Energy Econ. 127, 107059. doi:10.1016/j.eneco.2023.107059

CrossRef Full Text | Google Scholar

Meadows, D. H., Meadows, D. L., Randers, J., and Behrens III, W. W. (1972). The limits to growth: a report for the club of Rome’s project on the predicament of mankind. Universe Books. Available online at: https://www.clubofrome.org/publication/the-limits-to-growth/.

Google Scholar

Nam, K., Heo, S., Loy-Benitez, J., Ifaei, P., and Yoo, C. (2020). An autonomous operational trajectory searching system for an economic and environmental membrane bioreactor plant using deep reinforcement learning. Water Sci. Technol. 81 (8), 1578–1587. doi:10.2166/wst.2020.053

PubMed Abstract | CrossRef Full Text | Google Scholar

Osman, A., Chen, L., Yang, M., Msigwa, G., Farghali, M., Fawzy, S., et al. (2022). Cost, environmental impact, and resilience of renewable energy under a changing climate: a review. Environ. Chem. Lett. 21 (2), 741–764. doi:10.1007/s10311-022-01532-8

CrossRef Full Text | Google Scholar

Pasupuleti, M. (2025). GIS-driven AI for climate-resilient data centre siting and operation. Int. J. Acad. Industrial Res. Innovations 5 (11), 93–118. doi:10.62311/nesx/rp-nov-m5-2025

CrossRef Full Text | Google Scholar

Piemontese, L., Terzi, S., Baldassarre, G., Schwieger, D., Castelli, G., and Bresci, E. (2024). Over-reliance on water infrastructure can hinder climate resilience in pastoral drylands. Nat. Clim. Change 14 (3), 267–274. doi:10.1038/s41558-024-01929-z

CrossRef Full Text | Google Scholar

Pimenow, S., Pimenowa, O., and Prus, P. (2024). Challenges of artificial intelligence development in the context of energy consumption and impact on climate change. Energies 17 (23), 5965. doi:10.3390/en17235965

CrossRef Full Text | Google Scholar

Rane, N., Choudhary, S., and Rane, J. (2024). Artificial intelligence for enhancing resilience. J. Appl. Artif. Intell. 5 (2), 1–33. doi:10.48185/jaai.v5i2.1053

CrossRef Full Text | Google Scholar

Saeed, F., Schleussner, C. F., and Almazroui, M. (2021). From paris to Makkah: heat stress risks for Muslim pilgrims at 1.5 C and 2 C. Environ. Res. Lett. 16 (2), 024037. doi:10.1088/1748-9326/abd067

CrossRef Full Text | Google Scholar

Saldanha, J. J., Nied, A., Trentini, R., and Kutzner, R. (2024). AI-based optimal allocation of BESS, EV charging station and DG in distribution network for losses reduction and peak load shaving. Electr. Power Syst. Res. 234, 110554. doi:10.1016/j.epsr.2024.110554

CrossRef Full Text | Google Scholar

Satterthwaite, D., Huq, S., Reid, H., Pelling, M., and Lankao, P. R. (2012). “Adapting to climate change in urban areas: the possibilities and constraints in low-and middle-income nations 1,” in Adapting cities to climate change (London, New York: Routledge), 3–47.

Google Scholar

Schwartzman, D. (2025). Science-based ecosocialist strategy for climate security. J. World-Systems Res. 31 (1), 108–135. doi:10.5195/jwsr.2025.1323

CrossRef Full Text | Google Scholar

Şerban, A. C., and Lytras, M. D. (2020). Artificial intelligence for smart renewable energy sector in Europe—Smart energy infrastructures for next generation smart cities. IEEE Access 8, 77364–77377. doi:10.1109/ACCESS.2020.2990123

CrossRef Full Text | Google Scholar

Shahbaz, M., Wang, J., Dong, K., and Zhao, J. (2022). The impact of digital economy on energy transition across the globe: the mediating role of government governance. Renew. Sustain. Energy Rev. 166, 112620. doi:10.1016/j.rser.2022.112620

CrossRef Full Text | Google Scholar

Shen, Y., Shi, X., Zhao, Z., Grafton, R. Q., Yu, J., and Shan, Y. (2024). Quantifying energy transition vulnerability helps more just and inclusive decarbonization. PNAS Nexus 3 (10), pgae427. doi:10.1093/pnasnexus/pgae427

PubMed Abstract | CrossRef Full Text | Google Scholar

Shi, X., Cheong, T. S., and Li, V. J. (2021). Evolution of future world coal consumption: insights from a distribution dynamics approach. Int. J. Oil, Gas Coal Technol. 27 (2), 186–207. doi:10.1504/ijogct.2021.115546

CrossRef Full Text | Google Scholar

Strubell, E., Ganesh, A., and McCallum, A. (2019). “Energy and policy considerations for deep learning in NLP,” in Proceedings of the 57th annual meeting of the association for computational linguistics, 3645–3650. doi:10.18653/v1/P19-1355

CrossRef Full Text | Google Scholar

Summers, J. K., Smith, L. M., Harwell, L. C., and Buck, K. D. (2017). Conceptualizing holistic community resilience to climate events: foundation for a climate resilience screening index. GeoHealth 1 (4), 151–164.

PubMed Abstract | CrossRef Full Text | Google Scholar

Tabor, D. P., Roch, L. M., Saikin, S. K., Kreisbeck, C., Sheberla, D., Montoya, J. H., et al. (2018). Accelerating the discovery of materials for clean energy in the era of smart automation. Nat. Rev. Mater. 3 (5), 5–20. doi:10.1038/s41578-018-0005-z

CrossRef Full Text | Google Scholar

Treloar, G. J., Love, P. E. D., Faniran, O. O., and Iyer-Raniga, U. (2000). A hybrid life cycle assessment method for construction. Constr. Manag. and Econ. 18 (1), 5–9. doi:10.1080/014461900370898

CrossRef Full Text | Google Scholar

Tyler, S., and Moench, M. (2012). A framework for urban climate resilience. Clim. Dev. 4 (4), 311–326. doi:10.1080/17565529.2012.745389

CrossRef Full Text | Google Scholar

Tyler, S., Nugraha, E., Nguyen, H. K., Van Nguyen, N., Sari, A. D., Thinpanga, P., et al. (2016). Indicators of urban climate resilience: a contextual approach. Environ. Sci. and Policy 66, 420–426. doi:10.1016/j.envsci.2016.08.004

CrossRef Full Text | Google Scholar

Višković, A., Franki, V., and Jevtić, D. (2022). “Artificial intelligence as a facilitator of the energy transition,” in 2022 45th jubilee international convention on information, communication and electronic technology (MIPRO) (IEEE), 494–499. doi:10.23919/MIPRO55190.2022.9803700

CrossRef Full Text | Google Scholar

Waltersmann, L., Kiemel, S., Stuhlsatz, J., Sauer, A., and Miehe, R. (2021). Artificial intelligence applications for increasing resource efficiency in manufacturing companies—a comprehensive review. Sustainability 13 (12), 6689. doi:10.3390/su13126689

CrossRef Full Text | Google Scholar

Wang, D., and Chen, S. (2024). The effect of pilot climate-resilient city policies on urban climate resilience: evidence from quasi-natural experiments. Cities 153, 105316. doi:10.1016/j.cities.2024.105316

CrossRef Full Text | Google Scholar

Wang, Q., Hu, S., and Li, R. (2024a). Could information and communication technology (ICT) reduce carbon emissions? The role of trade openness and financial development. Telecommun. Policy 48 (3), 102699. doi:10.1016/j.telpol.2023.102699

CrossRef Full Text | Google Scholar

Wang, Q., Li, Y., and Li, R. (2024b). Ecological footprints, carbon emissions, and energy transitions: the impact of artificial intelligence (AI). Humanit. Soc. Sci. Commun. 11 (1), 1–18. doi:10.1057/s41599-024-03520-5

CrossRef Full Text | Google Scholar

Wang, Q., Zhang, F., Li, R., and Sun, J. (2024c). Does artificial intelligence promote energy transition and curb carbon emissions? The role of trade openness. J. Clean. Prod. 447, 141298. doi:10.30574/ijsra.2024.11.1.0112

CrossRef Full Text | Google Scholar

Wang, Y., Su, Z., Cai, X., and Yu, J. (2025). The dual carbon emission effects of digital economy: evidence from China. Heliyon 11, e42554. doi:10.1016/j.heliyon.2025.e42554

PubMed Abstract | CrossRef Full Text | Google Scholar

Wei, X., Zhang, P., and Du, Y. (2020). How robots reshape the urban labor market: from a perspective of migrants’ job tasks. Econ. Perspect. (10), 92–109. Available online at: https://jjxdt.ajcass.com/?jumpnotice=201903080001#/issue?id=76179&year=2025&issue=12.

Google Scholar

Xu, X., Yu, J., Zhang, D., and Ji, Q. (2021). Energy insecurity, economic growth and the role of renewable energy: a cross-country panel analysis. Singap. Econ. Rev. 66 (2), 323–343. doi:10.1142/S021759081943001X

CrossRef Full Text | Google Scholar

Xu, J., Chen, Y., Yang, N., and Shao, S. (2025). The impact of artificial intelligence on the energy transition: evidence from Chinese cities. World Dev. 195, 107126. doi:10.1016/j.worlddev.2025.107126

CrossRef Full Text | Google Scholar

Yu, J., Shi, X., Guo, D., and Yang, L. (2021). Economic policy uncertainty (EPU) and firm carbon emissions: evidence using a China provincial EPU index. Energy Econ. 94, 105071. doi:10.1016/j.eneco.2020.105071

CrossRef Full Text | Google Scholar

Yu, J., Liu, P., Fu, D., and Shi, X. (2023). How do power shortages affect CO2 emission intensity? firm-level evidence from China. Energy 282, 128927. doi:10.1016/j.energy.2023.128927

CrossRef Full Text | Google Scholar

Zavyalova, E. B., Volokhina, V. A., Troyanskaya, M. A., and Dubova, Y. I. (2023). A humanistic model of corporate social responsibility in e-commerce with high-tech support in the artificial intelligence economy. Humanit. Soc. Sci. Commun. 10 (1), 1–10. doi:10.1057/s41599-023-01764-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhao, Q., Wang, L., Stan, S. E., and Mirza, N. (2024). Can artificial intelligence help accelerate the transition to renewable energy? Energy Econ. 134, 107584. doi:10.1016/j.eneco.2024.107584

CrossRef Full Text | Google Scholar

Zhao, C., Yu, J., Li, P., and Shen, Y. (2025). The impact and spatial externalities of unstable power supply on the low-carbon transition in China. Energy Econ. 144, 108306. doi:10.1016/j.eneco.2025.108306

CrossRef Full Text | Google Scholar

Zhong, Y., and Li, Y. (2023). Statistical evaluation of sustainable urban planning: integrating renewable energy sources, energy-efficient buildings, and climate resilience measures. Sustain. Cities Soc. 101, 105160. doi:10.1016/j.scs.2023.105160

CrossRef Full Text | Google Scholar

Zhou, Y., and Liu, J. (2024). Advances in emerging digital technologies for energy efficiency and energy integration in smart cities. Energy Build. 315, 114289. doi:10.1016/j.enbuild.2024.114289

CrossRef Full Text | Google Scholar