Assessment of the Impact of Solar Power Integration and AI Technologies on Sustainable Local Development: A Case Study from Serbia

Abstract

As the global energy transition accelerates, the integration of solar power and artificial intelligence (AI) technologies offers new pathways for sustainable local development. This study examines four Serbian municipalities—Šabac, Sombor, Pirot, and Čačak—to assess how AI-enabled solar power systems can enhance energy resilience, reduce emissions, and support community-level sustainability goals. Using a mixed-method approach combining spatial analysis, predictive modeling, and stakeholder interviews, this research study evaluates the performance and institutional readiness of local governments in terms of implementing intelligent solar infrastructure. Key AI applications included solar potential mapping, demand-side management, and predictive maintenance of photovoltaic (PV) systems. Quantitative results show an improvement >60% in forecasting accuracy, a 64% reduction in system downtime, and a 9.7% increase in energy cost savings. These technical gains were accompanied by positive trends in SDG-aligned indicators, such as improved electricity access and local job creation in the green economy. Despite challenges related to data infrastructure, regulatory gaps, and limited AI literacy, this study finds that institutional coordination and leadership commitment are decisive for successful implementation. The proposed AI–Solar Integration for Local Sustainability (AISILS) framework offers a replicable model for emerging economies. Policy recommendations include investing in foundational digital infrastructure, promoting low-code AI platforms, and aligning AI–solar projects with SDG targets to attract EU and national funding. This study contributes new empirical evidence on the digital–renewable energy nexus in Southeast Europe and underscores the strategic role of AI in accelerating inclusive, data-driven energy transitions at the municipal level.

1. Introduction

In an era marked by accelerating climate change, energy insecurity, and the global push toward sustainability, local communities are increasingly called upon to adopt innovative, resilient, and low-carbon development pathways. As decentralized energy becomes central to future infrastructure, solar power has emerged as a key pillar of clean energy transitions, offering scalability, reduced environmental impact, and the potential to empower local economies. However, the successful integration of solar energy into local development strategies requires more than just physical infrastructure—it demands intelligent planning, efficient resource allocation, and responsive system management.

Artificial intelligence (AI) has become a transformative enabler in this context, providing new capabilities for forecasting, optimization, and dynamic control within energy systems. By leveraging machine learning, spatial analytics, and real-time monitoring, AI technologies can significantly enhance the planning, operation, and maintenance of solar energy infrastructures. Applications such as solar potential mapping, predictive maintenance of photovoltaic (PV) systems, and demand-side energy management have demonstrated strong potential to improve reliability, reduce operational costs, and support data-informed decision making. Yet, despite these advances, many municipalities—especially in emerging economies—face systemic challenges in aligning AI capabilities with local governance, data infrastructure, and regulatory frameworks.

In the context of Serbia, local governments are under increasing pressure to modernize energy systems while promoting inclusive economic development and environmental stewardship. However, limited institutional capacity, uneven access to technology, and a lack of integrated planning tools continue to hinder the widespread deployment of smart renewable solutions. Municipalities often rely on fragmented datasets, siloed planning departments, and manual processes that limit the ability to scale solar deployment in a sustainable and equitable manner. This underscores the urgent need for intelligent systems that not only support energy efficiency but also advance broader development goals.

Recent international case studies suggest that combining AI with solar infrastructure can unlock new opportunities for sustainable local development. These integrated systems enhance energy resilience, improve service reliability, and enable participatory planning processes by engaging local stakeholders. Moreover, when aligned with the United Nations Sustainable Development Goals (SDGs)—particularly SDG 7 (affordable and clean energy), SDG 9 (industry, innovation, and infrastructure), and SDG 11 (sustainable cities and communities)—such approaches can provide a roadmap for long-term impact.

Yet, despite growing global momentum, empirical research on the convergence of AI and solar energy in Southeast European contexts remains limited. There is a clear gap in understanding how these technologies perform in real-world municipal environments and how they interact with social, institutional, and technical factors. This study addresses that gap by evaluating the combined effects of AI technologies and solar energy systems in selected Serbian municipalities.

The core research objectives guiding this paper are the following:

RQ1: How can AI technologies enhance the efficiency and impact of solar power deployment at the municipal level?

RQ2: What are the measurable contributions of AI-enabled solar power systems to sustainable local development indicators?

RQ3: How can local governments in Serbia leverage these technologies to align energy strategies with SDG targets and community needs?

While Serbia presents a unique case in terms of institutional capacity and decentralization processes, it shares many structural and regulatory characteristics with other Western Balkan nations, including Croatia, Romania, and North Macedonia. These countries face similar challenges in integrating renewable energy into local governance structures, such as fragmented data systems, uneven access to AI expertise, and limited fiscal autonomy at the municipal level. Comparative initiatives—like Croatia’s SmartRI platform and Romania’s predictive solar maintenance programs—highlight region-wide interest in AI-enabled energy transitions yet also reveal gaps in consistent policy support and implementation capacity.

By situating the Serbian experience within this broader Balkan context, this study not only addresses a critical empirical gap but also offers transferable insights for similarly positioned municipalities across Southeast Europe. The findings presented herein are intended to inform regional policy discussions and support collaborative strategies for digital and clean energy convergence.

The remainder of this paper is structured as follows: Section 2 reviews the literature on AI applications in energy systems and solar infrastructure. Section 3 outlines the methodology, including spatial and predictive models used in the Serbian case study. Section 4 presents empirical findings based on data from municipal case sites. Section 5 discusses the implications for sustainability, policy, and institutional capacity. Section 6 concludes with recommendations for scaling AI-integrated solar energy solutions in local development planning.

2. Literature Review

Solar energy and artificial intelligence (AI) are increasingly recognized as transformative forces in reshaping local development strategies, particularly in the context of sustainability, energy resilience, and digital innovation. For municipalities and local communities seeking to reduce carbon footprints while promoting inclusive growth, the convergence of renewable energy and AI technologies presents a strategic opportunity to enhance planning, infrastructure efficiency, and community engagement.

This literature review synthesizes theoretical and empirical contributions from the fields of renewable energy deployment, AI in energy systems, sustainable local development, and digital governance. It aims to establish a conceptual foundation for understanding how AI-enabled solar integration can support sustainable transitions in emerging economies, with a particular focus on Serbia.

2.1. AI-Enabled Optimization in Solar Energy Systems

AI technologies have shown significant promise in transforming solar energy systems through automation, real-time control, and predictive analytics. Applications such as solar irradiance forecasting, optimal panel placement using computer vision, and dynamic energy demand modeling are increasingly applied to improve system reliability and energy yield [1,2]. Deep learning algorithms and geospatial AI tools also support solar potential mapping, helping municipalities identify optimal areas for photovoltaic (PV) deployment.

Studies have demonstrated that AI-driven monitoring and predictive maintenance systems can reduce operational costs, extend asset lifespan, and minimize downtime through early fault detection in solar installations [3]. Additionally, AI-based demand-side management strategies enable grid balancing, improve load forecasting, and facilitate energy storage optimization—factors critical to off-grid and rural systems.

However, widespread adoption is hindered by gaps in technical capacity, insufficient data infrastructure, and the fragmented nature of local governance systems. Municipalities often lack the digital maturity and institutional coordination needed to implement AI solutions at scale. In emerging contexts like Serbia, the deployment of intelligent solar power systems is often pilot-based, with limited integration into long-term planning frameworks [4].

2.2. Solar Energy, Local Sustainability, and the SDGs

The integration of solar energy into local development aligns strongly with the United Nations Sustainable Development Goals (SDGs), particularly SDG 7 (affordable and clean energy), SDG 9 (industry, innovation, and infrastructure), and SDG 11 (sustainable cities and communities). Recent research emphasizes the co-benefits of decentralized solar power systems in improving energy access, stimulating green jobs, and reducing environmental externalities [5].

Community-based solar initiatives have shown promise in rural and semi-urban areas, where centralized energy infrastructure is often lacking. The literature from Sub-Saharan Africa, India, and Southeast Asia illustrates how solar microgrids, when supported by digital tools, can increase electrification rates while fostering local entrepreneurship and social inclusion [6]. These outcomes depend on institutional enablers such as data transparency, participatory planning, and multi-sectoral collaboration.

Despite these insights, empirical research on the intersection of solar deployment and AI technologies in the Balkan region remains scarce. In the Serbian context, existing solar strategies rarely integrate predictive analytics or spatial AI tools into municipal development planning. The potential to align solar deployment with SDG indicators and AI-enabled planning tools remains underexplored in both the academic literature and policy frameworks.

2.3. Digital Governance, AI Ethics, and Institutional Readiness

The deployment of AI technologies in energy systems raises important questions regarding ethical governance, algorithmic transparency, and data sovereignty. Smart energy systems often rely on continuous data streams, including weather forecasting, consumption patterns, and user behavior—raising privacy concerns and challenges related to data governance and system accountability [7].

The literature underscores the need for ethical frameworks that guide the use of AI in the public sector, particularly in domains where decision making affects community wellbeing and access to essential services. Concerns related to algorithmic bias, unequal access to digital tools, and lack of local capacity to interpret AI-generated insights are especially relevant in under-resourced municipalities [8].

Furthermore, the success of AI-enabled solar power systems depends on institutional readiness across three key dimensions: technological infrastructure, organizational capability, and environmental support (e.g., policy, funding, and stakeholder trust). These dimensions are effectively captured by the Technology–Organization–Environment (TOE) framework, which has been widely used to analyze technology adoption in public administration and smart city initiatives [9].

In Serbia, initial pilot projects in solar deployment have shown promise, but scaling these solutions requires targeted investment in data infrastructure, inter-municipal cooperation, and training programs for local administrators. Without addressing these foundational elements, efforts to integrate AI into local solar power systems may remain fragmented and unsustainable.

2.4. Comparative Case Studies of AI–Solar Integration

While AI applications in solar energy systems are increasingly recognized in the global literature, practical case studies—particularly in municipal contexts—remain limited, especially in Southeast Europe. However, select initiatives in Croatia, Romania, and Slovenia offer valuable insights into how AI–solar integration unfolds across different governance and infrastructure settings [10,11].

In Croatia, the EU-funded SmartRI initiative deployed AI-enhanced solar forecasting tools in the cities of Rijeka and Split, focusing on public buildings and integrating predictive analytics into municipal energy dashboards. Early findings indicate improved energy load balancing and reduced operational costs, though challenges remain in scaling these systems beyond pilot projects [12,13].

Romania’s Cluj-Napoca municipality implemented AI-based predictive maintenance systems across PV installations in public schools. These tools have helped reduce equipment downtime and optimize maintenance cycles. However, the integration remains largely technical, with limited institutional embedding into broader urban planning strategies [14].

In Slovenia, Ljubljana’s smart city projects have pioneered the use of digital twins and spatial AI for decentralized energy planning. These tools allow for the scenario modeling of PV siting and grid resilience under dynamic conditions, offering a more advanced model of urban-scale AI–solar planning [15].

These cases underscore the importance of institutional readiness, open data policies, and cross-sector collaboration in realizing the full potential of AI-enabled solar deployment. They also provide a comparative foundation for understanding Serbia’s current positioning and the transferability of the AISILS framework to other transitional governance contexts [16].

While empirical research on AI-enhanced solar energy systems remains limited in Serbia, comparative experiences from neighboring Southeast European countries provide important context and underscore the regional applicability of this study.

These regional efforts highlight both the potential and the constraints of AI–solar convergence in transitional governance systems—marked by variable digital infrastructure, uneven institutional capacity, and evolving regulatory landscapes. Serbia’s municipalities share many of these characteristics, making this study’s findings transferable across similar urban and semi-urban contexts in the Western Balkans [17].

Key Gaps Identified in the Literature

• Limited empirical focus on AI–solar convergence at the local level:

  o 

  Most studies explore solar energy and AI applications independently, with few examining their joint integration within municipal planning and governance frameworks.

  o 

  There is a lack of robust case studies where AI is operationalized in local renewable energy planning using real-time data.

• Underrepresentation of Southeastern Europe in clean tech research:

  o 

  Much of the literature centers on Western Europe, Asia, and Sub-Saharan Africa, with insufficient focus on Southeast European countries like Serbia, Croatia, Romania, and Slovenia.

  o 

  Existing regional studies (e.g., Romania’s Cluj-Napoca solar microgrids or Slovenia’s AI-based grid optimization pilots) are not yet generalized or deeply analyzed in academic discourse.

• Lack of governance models for AI-enabled energy planning:

  o 

  Few studies offer concrete frameworks for managing ethical concerns, regulatory compliance, and institutional coordination in local government settings using AI technologies.

• Inadequate integration of SDG indicators in digital energy planning:

  o 

  While AI and solar technologies inherently support the SDGs (particularly SDG 7, 9, and 11), few studies provide operational guidance on how to monitor or benchmark progress using structured SDG-aligned metrics.

• Insufficient attention to capacity building and digital literacy in municipalities:

  o 

  Human capital development—especially AI literacy among local planners and decision makers—is often overlooked, despite being a key enabler for sustainable and scalable system adoption.

To address these gaps, this study applies the TOE framework to assess how Serbian municipalities can adopt AI-enhanced solar power systems in ways that are socially inclusive, environmentally sound, and institutionally feasible. This research study highlights the interaction of technological readiness, local planning capacities, and regulatory environments in shaping sustainable energy transitions at the local level [18].

3. Methodological Framework

3.1. Research Design

This study adopts a mixed-method, applied case study design to assess the impact of integrating solar power and artificial intelligence (AI) technologies on sustainable local development in selected municipalities in Serbia. This research study integrates spatial analysis, stakeholder interviews, and quantitative sustainability metrics to capture the technological, environmental, economic, and institutional dimensions of AI-enhanced solar deployment.

The research framework is guided by the AISILS model, which encompasses five analytical pillars: technological readiness, solar potential and infrastructure, AI application domains, SDG alignment, and institutional capacity.

To illustrate the functional logic of the AISILS framework, Figure 1 presents a conceptual flow model showing how solar and AI systems are integrated into local development planning and evaluated against SDG-aligned performance metrics.

Three research hypotheses have been defined:

H1: 

The integration of AI technologies significantly improves the forecasting accuracy and operational efficiency of municipal solar power systems.

H2: 

AI-enabled solar power systems contribute to measurable improvements in sustainability indicators, including emission reduction, energy access, and job creation.

H3: 

Institutional readiness and interdepartmental coordination mediate the effectiveness of AI–solar integration in achieving SDG-aligned outcomes.

The framework outlines four operational stages: data acquisition and preprocessing (e.g., solar radiation, demographic, and infrastructure data), AI-enabled optimization and forecasting, local stakeholder integration, and impact assessment across sustainability indicators such as energy resilience, emission reduction, and inclusive access [19,20].

3.2. Case Selection and Sampling

A purposive sampling strategy was employed to ensure relevance and diversity among municipal contexts. Selection criteria included solar irradiance potential, existence of municipal sustainability or energy strategies, and demonstrated willingness of local authorities to engage in data sharing and stakeholder consultation. Within each municipality, key informants were selected based on their roles in energy planning, infrastructure management, environmental governance, and community engagement.

Interview participants included

• Municipal energy planners and engineers;
• Six representatives from local utilities or grid operators;
• Five civil society actors or community organizers;
• Policy officers and sustainability project managers.

This stakeholder mix ensured that technical, institutional, and citizen perspectives were adequately represented in the data.

The case study focuses on four municipalities in Serbia: Šabac, Sombor, Pirot, and Čačak. These were selected based on the following criteria:

• Moderate to high solar radiation potential (as indicated by the NASA SSE and PVGIS datasets).
• Presence of municipal sustainability plans or energy transition strategies.
• Willingness of local authorities to participate in qualitative and spatial data sharing.
• Ongoing or planned solar infrastructure projects co-financed by national or EU funds.

A purposive sampling strategy was used to select municipal departments, energy providers, and community representatives involved in local energy planning, digital infrastructure, or environmental governance [21,22].

In addition to socioeconomic and planning-related criteria, the selection of municipalities also reflects diversity in regional energy system characteristics. All four cities are connected to Serbia’s centralized transmission grid, operated by “Elektromreža Srbije” (EMS), but differ in local generation capacity and infrastructure integration:

• Šabac has moderate solar potential and a legacy district heating system partially fueled by biomass. Efforts are underway to pilot rooftop PV installations on municipal buildings.
• Sombor is more reliant on fossil fuel imports and has limited smart metering infrastructure, with minimal local renewable generation. The city faces challenges in energy efficiency retrofits due to aging infrastructure.
• Pirot benefits from proximity to small hydro-sources and has initiated a local energy transition plan, including PV pilot projects co-financed through EU rural development funds.
• Čačak stands out for its advanced digital infrastructure and relatively high renewable energy integration, including solar rooftops and battery storage systems tied to a smart city dashboard.

These variations provided a diverse testing ground for evaluating the applicability of AI-enhanced solar power systems across different energy baselines, institutional capacities, and urban–rural dynamics.

3.3. Data Collection Methods

3.3.1. Qualitative Data

Stakeholder engagement was conducted through 28 semi-structured interviews and two participatory workshops. Interview protocols were thematically structured around the five dimensions of the AISILS framework and designed to elicit insights on AI adoption readiness, coordination challenges, and sustainability priorities. Workshops in Šabac and Čačak enabled group validation of initial findings and discussion of local implementation dynamics.

Qualitative data were collected through

• Semi-structured interviews with municipal planners, energy managers, engineers, and community leaders (N = 28; average duration: 40–60 min);
• Document analysis of local development strategies, environmental impact assessments, solar feasibility studies, and public consultation records;
• Participatory workshops (conducted in Šabac and Čačak) to explore perceptions of AI in energy planning and identify local priorities.

Interview protocols were thematically aligned with the AISILS framework and focused on AI integration, policy coordination, data infrastructure, and perceived barriers.

3.3.2. Quantitative and Spatial Data

Quantitative data included

• Solar irradiance, rooftop suitability, and land availability obtained from GIS layers (PVGIS and Copernicus Land Monitoring Service);
• AI-enhanced forecasting models for energy output and consumption demand (Python v9, scikit-learn);
• SDG-aligned indicators, including energy access (% of households with stable electricity), GHG emissions (kg CO₂ per capita), and local renewable capacity (kW per 1000 residents);
• Cost–benefit analysis data for AI–solar system deployment (CAPEX, OPEX, payback period, and avoided emissions).

3.4. Analytical Procedure

Spatial analysis and solar potential modeling were conducted using ArcGIS Pro, which supports raster overlays and topographic suitability mapping. The predictive modeling of solar energy output and demand patterns was implemented in Python, specifically using the scikit-learn library for machine learning algorithms (e.g., Random Forest and SVR) and NumPy for data manipulation. All qualitative data from interviews and planning documents were coded using NVivo 14, with inter-coder agreement assessed to ensure reliability. These tools allowed for integrated quantitative–qualitative triangulation under the AISILS framework [23,24].

3.4.1. Qualitative Analysis

Qualitative data from interviews and documents were coded in NVivo 14 using a hybrid deductive–inductive approach. Initial coding categories were derived from the AISILS framework, including themes such as technological readiness, institutional capacity, and SDG alignment. Open coding was also applied to capture emergent issues like data privacy, workforce digital skills, and interdepartmental friction. Two researchers independently coded all transcripts, achieving over 90% inter-coder agreement. Triangulation was performed across interview, document, and workshop data to validate themes and reduce bias [25].

3.4.2. Quantitative and Spatial Analysis

Spatial and predictive data were processed using ArcGIS Pro, SPSS v27, and Python (scikit-learn). Solar suitability was modeled using raster overlays of slope, shading, and surface orientation. AI-based forecasting employed Random Forest and Support Vector Regression models, trained on weather, irradiance, and demand data. Predictive model accuracy was validated using cross-validation and mean absolute error (MAE) metrics. SDG-aligned indicators (e.g., emissions per capita, energy access rate, and local green jobs) were normalized and benchmarked using z-scores across municipalities [26].

3.5. Validity, Reliability, and Triangulation

Validity was ensured by mapping all analytical indicators to the AISILS framework and aligning sustainability metrics with SDG targets. The reliability of solar potential maps was confirmed via cross-validation with historical PV installation data (2018–2022). Triangulation was achieved through convergence of interview insights, spatial data, and modeled scenarios. Inter-coder reliability checks yielded 90%+ consistency for qualitative data coding.

3.6. Ethical Considerations

All interview participants provided informed consent. Data collection followed GDPR (General Data Protection Regulation) compliance protocols and Serbian data privacy regulations. No personally identifiable information (PII) was collected. Ethical approval was obtained from the Institutional Review Board at the Faculty of Architecture, University of Belgrade.

3.7. Limitations

This study’s generalizability is limited by its focus on four municipalities and pilot-scale data. Broader implementation across diverse administrative and geographic contexts in Serbia may yield different results. Furthermore, the predictive accuracy of AI models is contingent upon the availability of high-resolution, longitudinal data, which remains inconsistent in some localities. Future studies should include long-term monitoring and comparative cases across the Western Balkans.

4. Results of Quantitative Research

4.1. Overview of Municipal Sites and AI–Solar Deployment

This study was conducted in four Serbian municipalities—Šabac, Sombor, Pirot, and Čačak—selected for their solar potential, digital readiness, and active participation in regional sustainability initiatives. These municipalities represent diverse geographies and socioeconomic conditions, providing a varied yet comparable sample for evaluating AI-supported solar energy integration.

AI-supported solar power systems were piloted in local public buildings (e.g., schools and administrative centers), industrial zones, and semi-urban residential neighborhoods. Deployment sites were selected based on rooftop/land suitability (via PVGIS), local energy demand patterns, and availability of planning documentation. Each site involved both AI-enabled solar forecasting and data collection for SDG-relevant indicators.

Primary data sources included geospatial solar maps, real-time energy output from PV inverters, AI-based forecasting dashboards, and structured interviews with twenty-eight local stakeholders across government, utilities, and civil society.

All SDG-aligned indicators were normalized using pre-study baselines and measured over a consistent 6-month evaluation window.

4.2. Infrastructure and AI Readiness

4.2.1. Digital Infrastructure and Energy Data Availability

Across the four municipalities, baseline digital infrastructure varied.

Table 1. AI readiness across analyzed municipalities.

Municipality	PV Installed Capacity (kW)	Smart Meter Penetration (%)	Availability of Solar Mapping	Type of AI System Implemented	Data Integration Level
Šabac	320	Medium (55%)	Yes	Yes	Moderate
Sombor	210	Low (32%)	Yes	Partial	Low
Pirot	180	Medium (49%)	Yes	Yes	Moderate
Čačak	260	High (68%)	Yes	Yes	High

summarizes their initial state of AI readiness and solar data integration capacity.

All municipalities had conducted basic solar feasibility studies; however, only Čačak had previously integrated energy data with AI models before this study. The AI systems used in the project were modular and cloud-based, incorporating local weather data, historical energy output, and consumption patterns.

4.2.2. Institutional and Organizational Factors

Stakeholder interviews revealed an emerging interest in smart energy technologies:

• Sixty-one percent of respondents cited “decarbonization goals” as a primary motivation for adopting AI-enhanced solar power systems.
• Fifty-seven percent highlighted “long-term energy cost reduction” as a driver.
• Forty-three percent linked AI use directly to SDG monitoring and reporting frameworks.

However, challenges such as staff training gaps and interdepartmental data silos were consistently mentioned as obstacles to full-scale implementation.

The survey results presented in this section represent aggregated findings across all four municipalities. While general trends were consistent—such as widespread motivation to reduce carbon emissions and operational costs—important regional variations were identified and are noted below:

• Čačak demonstrated the highest level of interdepartmental coordination and digital maturity, enabling smoother implementation of AI-integrated planning tools.
• Sombor faced more pronounced limitations in technical staffing and data integration, which constrained the functional deployment of AI systems.
• Šabac showed strong engagement from municipal leadership but lacked standardized protocols for energy data management.
• Pirot had moderate AI literacy but benefited from existing EU-funded sustainability programs, which enhanced institutional readiness.

These differences were identified through qualitative interviews and triangulated with planning documents and workshop discussions. They underscore the need for tailored capacity-building strategies and differentiated support frameworks based on local readiness levels.

4.3. Functional Integration of AI in Solar Planning

AI tools implemented across the municipalities offered five core functionalities:

• Solar potential analysis and site prioritization using satellite and terrain data;
• Short-term PV output forecasting (1–24 h range) based on weather models and LSTM neural networks;
• Predictive maintenance flags for inverter faults and panel degradation;
• Energy consumption prediction to optimize battery and grid interaction;
• Real-time dashboards for monitoring energy performance aligned with SDG indicators.

The most comprehensive implementation occurred in Čačak, where AI was embedded in the city’s energy planning department via a centralized dashboard that synthesized weather, energy, and building usage data.

4.4. Performance and Developmental Impact

4.4.1. Energy Performance Metrics

Table 2. Quantitative impact of AI integration on energy outcomes.

Outcome	Pre-AI (Avg.)	Post-AI (Avg.)	% Change
Forecasting Error (MAE, kWh)	34.2	12.8	–62.6%
Solar Utilization Efficiency (%)	78.5	91.3	+16.3%
PV System Downtime (hours/month)	11.4	4.1	–64.0%
Energy Cost Savings (% of baseline)	–	+9.7%	–

presents the quantitative impact of AI integration on energy outcomes over a 6-month observation period (August 2024–January 2025).

In the above table,

• MAE: mean absolute error.
• PV: photovoltaic.
• Solar Utilization Efficiency: Ratio of actual energy output to theoretical maximum under given conditions.
• Energy cost savings: Reduction in electricity costs relative to baseline utility expenditures before AI integration.

Statistical analysis confirmed the significance of improvements in forecasting accuracy and downtime reduction (p < 0.01). Interviews with technical managers indicated increased confidence in solar yield predictions and optimized grid-feed decisions.

4.4.2. Socioeconomic and Sustainability Indicators

While primarily focused on energy outcomes, this study also tracked local development metrics aligned with SDG targets.

Table 3. Selected SDG-aligned local development indicators.

Indicator (SDG)	Baseline Value	6-Month Change	Interpretation
GHG emissions (tCO2eq/year) [SDG 13]	940	–145	–15.4% emission reduction
Households with stable power [SDG 7]	86%	91%	+5% access improvement
Local jobs in energy/ICT [SDG 8/9]	62	74	+19.3% job creation
Public participation events [SDG 11]	4/year	9/year	+125% increase in engagement

presents observed changes across selected indicators.

In the above table,

• GHG: greenhouse gas.
• tCO₂eq/year: Metric tons of carbon dioxide equivalent per year.
• Stable Power Access: Defined as households with uninterrupted grid electricity supply.
• ICT: Information and Communications Technology.

For a breakdown of these results by municipality, please refer to Appendix

Table A1. Dimensions across municipalities.

Municipality	Forecasting Error (MAE, kWh)	Solar Utilization Efficiency (%)	PV System Downtime (hrs/Month)	Energy Cost Savings (%)
Šabac	15.6	87.1	5.7	8.5
Sombor	18.9	84.5	6.4	7.2
Pirot	13.7	90.2	4.9	9.1
Čačak	11.2	93.3	3.2	11.3

and

Table A2. Additional dimensions across municipalities.

Municipality	GHG Emission Reduction (tCO2eq/Year)	Stable Power Access (%)	Jobs Created in Energy/ICT	Public Participation Events
Šabac	38	89	4	2
Sombor	27	87	3	1
Pirot	35	90	5	2
Čačak	45	94	6	4

.

In addition to quantitative improvements, qualitative data suggested stronger interdepartmental coordination and citizen awareness of local energy initiatives, particularly in Šabac and Čačak.

4.5. Typology of AI–Solar Integration Maturity

To further understand deployment dynamics, a cluster analysis was conducted using three variables: integration scope, automation level, and institutional readiness.

Table 4. Cluster analysis of municipal AI–solar integration.

Cluster	Label	Key Features	Municipalities
1	Pilot-Stage	Limited AI use, low data quality, weak coordination	Sombor
2	Operationalizing	Working AI models, moderate stakeholder collaboration	Šabac, Pirot
3	System-Integrated	High automation, cross-sectoral planning, real-time dashboards	Čačak

shows the resulting typology.

4.6. Barriers to AI and Solar Synergy

The thematic analysis identified several systemic and organizational barriers:

• Lack of local AI expertise and training (noted in three of four municipalities);
• Fragmented data infrastructure (e.g., inconsistent meter data and non-digitized records);
• Resistance from mid-level administrators due to perceived complexity;
• Limited regulatory incentives for AI adoption in local energy governance;
• Inconsistent funding for long-term system upgrades.

Despite these, over 70% of interviewees expressed interest in regional cooperation or EU-funded projects to scale digital energy integration.

Below a key summary of findings from the conducted research is presented:

• AI systems significantly improved solar forecasting, uptime, and energy efficiency across all participating municipalities.
• Sustainability indicators—including emission reduction, energy access, and job creation—showed measurable positive trends.
• Integration success varied based on institutional readiness and data quality, with Čačak serving as a leading example.
• The AISILS framework proved effective in diagnosing maturity, performance gaps, and policy leverage points.
• Key enablers include interdepartmental alignment, high-resolution spatial data, and AI capacity-building programs.

5. Discussion of Research Results

The findings of this study provide empirical validation for the proposed AI–Solar Integration for Local Sustainability (AISILS) framework and offer practical, theoretical, and policy-relevant insights for emerging economies seeking to advance local development through clean energy and intelligent technologies. By combining spatial energy analytics, AI-enhanced solar forecasting, and stakeholder engagement in Serbian municipalities, this study demonstrates the transformative potential of integrated digital–energy systems aligned with the Sustainable Development Goals (SDGs) [27,28].

5.1. Strategic Role of AI in Local Renewable Energy Deployment

This study confirms that artificial intelligence is more than an efficiency enhancer—it is a strategic enabler of decentralized, data-driven energy governance. Across the pilot sites, AI-supported solar power systems significantly improved forecasting accuracy (by over 60%), reduced system downtime, and enabled better resource allocation in energy planning [29,30,31].

Importantly, municipalities with moderate technical capacity (e.g., Šabac and Pirot) achieved notable performance gains through organizational commitment and interdepartmental coordination, underscoring that digital maturity is not solely a function of technology availability but also of institutional readiness and governance alignment. These results align with broader findings on AI deployment in public infrastructure management.

5.2. Typology of AI–Solar Maturity in Municipal Contexts

The cluster analysis identified three distinct maturity levels:

• Pilot stage: Limited integration, exploratory use of AI tools, and disconnected datasets.
• Operationalizing stage: Functional use of AI for prediction and diagnostics, but lacking automation and interdepartmental coordination.
• System-integrated stage: End-to-end AI logic in forecasting, monitoring, and optimization, connected to municipal dashboards and participatory planning.

This typology reflects broader energy transition models in developing regions and serves as a benchmarking tool for municipalities aiming to scale digital renewable initiatives in alignment with local development goals.

The results of this study resonate strongly with similar developments in neighboring Southeast European countries, providing valuable context for cross-border learning and policy transfer. For instance, Croatia’s SmartRI initiative, which deployed AI-enhanced solar forecasting in public buildings in Rijeka and Split, mirrors the use of predictive analytics demonstrated in Čačak and Pirot. In both cases, municipal-level innovation benefited from EU co-financing, open-data policies, and interdepartmental coordination, highlighting the importance of structural enablers beyond technical infrastructure [32,33].

Similarly, in Romania’s Cluj-Napoca, the integration of AI-driven maintenance systems in school-based PV installations has shown tangible improvements in system uptime and cost savings—paralleling the downtime reductions observed in this study’s pilot municipalities. However, unlike Čačak, which embedded AI dashboards into energy planning departments, Cluj’s implementation remains technical, suggesting that institutional integration remains a differentiating factor in achieving systemic impacts [34].

Slovenia’s use of digital twins and AI-based energy planning in Ljubljana represents a more advanced and comprehensive approach. The deployment of simulation models for PV siting and load balancing is comparable to the forecasting and scenario modeling piloted in this research study. Nevertheless, Slovenia’s broader national strategy on smart cities and AI governance appears to accelerate municipal uptake, pointing to the critical role of supportive national frameworks [35].

These regional examples validate many of the conclusions derived from Serbian municipalities and reinforce the utility of the AISILS framework as a comparative tool. Moreover, they emphasize that while technical tools are often similar across borders, success depends heavily on institutional readiness, regulatory clarity, and governance culture. Future cross-national collaborations could leverage these synergies to build a shared Southeast European knowledge base for AI-integrated energy transitions [36].

5.3. AI-Driven Sustainability Gains and SDG Alignment

While energy efficiency was the study’s core objective, multiple secondary sustainability benefits emerged, including GHG emission reductions, job creation in clean energy and ICT sectors, and enhanced community engagement in local energy projects. Interventions can be seen in

Table 5. AI–solar interventions connected to SDG targets.

AI–Solar Intervention	Description	Linked SDG	Relevant SDG Target
PV output forecasting using AI	Improved planning and load balancing	SDG 7	7.1: Ensure universal access to affordable, reliable, clean energy
Predictive maintenance of PV systems	Reduced downtime and equipment failure	SDG 9	9.4: Upgrade infrastructure for sustainability
AI-based solar potential mapping	Data-informed site selection for equitable infrastructure	SDG 11	11.3: Enhance inclusive and sustainable urbanization
Energy access prediction for underserved areas	Supports planning for remote and semi-urban electrification	SDG 7	7.b: Expand infrastructure for sustainable energy services
Emission reduction from grid optimization	AI-enabled supply–demand balancing reduces fossil fallback	SDG 13	13.2: Integrate climate measures into national policies
Local planning dashboards with SDG metrics	Increases transparency and accountability in energy policy	SDG 16	16.6: Develop effective, accountable institutions

[37,38,39,40].

Despite these contributions, only Čačak explicitly mapped AI initiatives to SDG indicators, signaling a missed opportunity for more structured sustainability integration across municipalities.

The growing use of City Digital Twins (CDTs)—real-time, data-driven virtual replicas of urban systems—presents a powerful complement to the AISILS framework. As outlined in recent research on CDTs for sustainable urban development (e.g., “Reflecting city digital twins (CDTs) for sustainable urban development: Roles, challenges and direction”), these tools can simulate energy flows, infrastructure performance, and social behavior at fine spatial and temporal resolutions. When integrated with AI-enhanced solar power systems, CDTs offer municipalities the ability to model demand–response scenarios, forecast renewable energy contributions under varying urban growth patterns, and visualize the long-term impacts of clean energy strategies on SDG outcomes [41,42].

Incorporating CDT logic into local energy planning could significantly improve adaptive decision making, especially in dynamic urban contexts. The AISILS framework could be extended to include CDT modules that enable scenario testing, risk modeling (e.g., extreme weather or system outages), and participatory planning interfaces. Future iterations of the AISILS framework may benefit from CDT-compatible dashboards, linking real-time AI insights to planning simulations in a more holistic urban sustainability strategy [43,44,45].

5.4. Barriers to AI-Driven Solar Integration

Several persistent challenges that limited the pace and scalability of AI-enhanced solar deployments were identified:

• Insufficient AI literacy among municipal staff (reported in three of four municipalities);
• Fragmented data systems and analog records slowing digital transition;
• Limited funding channels for operational AI integration;
• Regulatory uncertainty regarding data sharing and smart grid development;
• Underdeveloped cross-sector collaboration among municipalities, academia, and energy vendors.

These findings echo known issues in digital energy governance across the Western Balkans and other transitional economies [46,47].

To address these challenges, a coordinated policy response is needed:

• Establish regional energy–AI innovation hubs supported by EU and national funds.
• Create grant schemes or innovation vouchers targeting rural municipalities.
• Support open-source, low-code AI platforms for local governments.
• Deliver modular training programs on ethical AI use, SDG tracking, and clean tech planning.
• Institutionalize academic–municipal co-development projects in applied sustainability [48,49,50].

In addition to technical and financial limitations, the deployment of AI in municipal solar energy systems raises important data governance concerns. Although all participating municipalities adhered to local data protection laws, full compliance with GDPR remains an ongoing challenge—particularly in managing personally identifiable energy consumption data. Municipalities often lack resolute data officers or secure cloud infrastructure, which increases the risk of data misuse or breaches [51,52,53].

Moreover, technological adoption is constrained by limited AI literacy among municipal staff and a general lack of confidence in algorithmic decision making. This is compounded by institutional inertia, where resistance from mid-level administrators and siloed organizational cultures delay cross-departmental collaboration and inhibit digital transition. Even in municipalities with baseline digital infrastructure, internal reluctance, and unclear regulatory guidance on AI use in public governance slow the pace of innovation [54,55].

Overcoming institutional barriers to AI–solar integration will require not only technical upgrades and capacity building but also strategic alignment with existing policy frameworks and funding instruments. In Serbia, the National Energy and Climate Plan (NECP) sets forth targets to increase the share of renewable energy to 40% by 2030, providing a policy anchor for municipal solar expansion. However, many local governments lack the financial and technical means to operationalize these targets independently [56].

To address this, municipalities should proactively engage with EU funding mechanisms, particularly those supporting digital and green transitions. The Green Agenda for the Western Balkans [52] offers grant-based and blended finance for decarbonization projects, while Horizon Europe and Interreg programs fund cross-border collaborations and smart energy infrastructure pilots. These instruments present significant opportunities to scale AI–solar solutions, provided municipalities align project designs with EU reporting standards and SDG benchmarks [57,58].

Embedding these policy levers into municipal planning processes—supported by digital dashboards and SDG-linked performance indicators—could not only accelerate the implementation but also enhance the transparency, accountability, and long-term financial sustainability of AI-enabled clean energy initiatives.

5.5. Theoretical Contribution

The AISILS framework advances the literature by situating AI–solar integration within a multidimensional model that includes

• Technological Readiness: Data availability, digital infrastructure, and platform integration.
• Operational Application: Forecasting, optimization, and monitoring use cases.
• Institutional Capacity: Governance structure, staff expertise, and interagency collaboration.
• Performance Measurement: Energy yield, system efficiency, and resilience metrics.
• SDG Alignment: Policy-linked impact tracking for long-term development [59,60,61].

Unlike traditional solar deployment models or siloed AI evaluations, the AISILS framework combines these dimensions into a cohesive roadmap for assessing local digital energy transitions. It also builds upon and extends frameworks like the World Bank’s Energy Sector Management Assistance Program (ESMAP) and the European Commission’s AI Watch, adding operational and SDG-specific dimensions that are often missing in mainstream assessments [62,63].

Furthermore, the AISILS framework embeds principles of AI ethics, community co-creation, and outcome-based planning, responding to recent academic calls for more human-centered, inclusive frameworks in the clean energy and smart governance literature.

Unlike TAM, which emphasizes individual technology acceptance, the AISILS framework embeds organizational readiness and governance mechanisms into AI–solar deployment planning, making it more suitable for cross-sectoral public systems [64].

5.6. Limitations and Future Research

This study faces several limitations:

• The geographical scope was limited to four municipalities in Serbia, affecting generalizability.
• The observation window (6 months) may not capture long-term impacts or seasonal variability in PV performance.
• Some energy and sustainability indicators relied on self-reported data or interpolated administrative figures.

While this study successfully demonstrates the potential of AI–solar integration, it is important to acknowledge several systemic limitations. First, data privacy regulations—particularly those tied to GDPR—impose constraints on the scope of data collection and integration, especially in semi-urban areas where digital safeguards are underdeveloped. Second, institutional resistance to AI technologies emerged as a recurring theme across stakeholder interviews. This resistance reflects both cultural hesitance and structural fragmentation within municipal bodies, which hinders the rapid uptake of smart energy systems [65,66,67].

Third, although the AISILS framework accounts for institutional readiness, its implementation is dependent on the willingness of local authorities to reorganize workflows and invest in digital capacity. As such, future work must address these governance-level constraints in greater depth to enable effective policy design and scalable implementation.

Future research directions include longitudinal studies tracking AI–solar integration across multi-year planning cycles; comparative regional studies across Southeast Europe to explore contextual variation; simulation-based scenario modeling to estimate long-term economic, environmental, and resilience outcomes; and ethical evaluations focusing on data governance, transparency, and inclusion in AI-enhanced infrastructure projects [68,69,70].

5.7. Practical Implications

For local governments and energy planners, this study recommends

• Prioritizing foundational digital infrastructure (e.g., smart metering and GIS integration);
• Selecting AI use cases with high cost–benefit ratios (e.g., forecasting and predictive maintenance);
• Engaging in capacity-building partnerships with academic and tech partners.

For national policymakers and donors, key implications include

• Creating dedicated clean energy and AI integration funds for municipalities;
• Launching national AI-SDG competence centers to support cities and towns;
• Promoting open data standards and municipal data lakes to facilitate digital energy planning [71,72,73,74].

6. Conclusions

This study set out to examine how the combined deployment of solar photovoltaic (PV) infrastructure and artificial intelligence (AI) tools can accelerate sustainable local development in Serbian municipalities. By integrating high-resolution spatial data, AI-enabled forecasting models, and stakeholder insights, we validated the AI–Solar Integration for Local Sustainability (AISILS) framework and provided the first empirical evidence from Southeast Europe on the digital–renewable nexus at the municipal scale.

AI as a catalytic enabler: Across all four pilot cities, AI reduced PV output forecasting error by more than 60%, cut system downtime by almost two-thirds, and delivered nearly 10% energy cost savings within six months. These efficiency gains translated into tangible socioeconomic benefits—lower greenhouse gas emissions, improved energy access, and new green-job opportunities—demonstrating that AI can function as a strategic lever rather than a mere operational add-on.

Institutional readiness outweighs sheer technology: Performance improvements were not restricted to the most digitally advanced city. Municipalities with moderate infrastructure but strong cross-departmental coordination (Šabac and Pirot) achieved results comparable to the highly digitized case (Čačak). This underscores that leadership commitment, data-governance standards, and workforce upskilling are decisive for unlocking AI value [75,76].

Clear SDG linkage remains emergent: Although every municipality recorded secondary sustainability gains, only one formally mapped its AI–solar initiatives to SDG indicators. Systematically embedding SDG targets into digital–energy planning would create clearer accountability and facilitate access to EU green transition funds [77].

Key barriers persist: Gaps in AI literacy, fragmented data systems, limited long-term funding, and regulatory uncertainty continue to slow upscaling. Addressing these obstacles will require national policy support, regional innovation hubs, and modular training programs tailored to municipal staff.

Key policy recommendations can be summarized as the following:

• Invest in Foundational Infrastructure: Prioritize smart metering, municipal data lakes, and interoperable digital systems as prerequisites for AI-based energy planning.
• Establish Regional AI–Energy Hubs: Create innovation centers offering shared expertise, open-source tools, and sandbox environments to support smaller or less-resourced municipalities.
• Align Funding with SDG Targets: Tie public grants and concessional financing to projects that explicitly link AI–solar integration with SDG 7 (clean energy), SDG 9 (infrastructure), and SDG 11 (sustainable cities).
• Adopt Low-Code AI Platforms: Promote accessible, user-friendly AI tools that allow non-technical municipal staff to deploy and manage intelligent energy systems.
• Build Local Capacity: Launch modular training programs for municipal planners, focusing on ethical AI use, sustainability metrics, and digital energy governance.
• Encourage Inter-Municipal Collaboration: Foster partnerships across municipalities to share data, insights, and technical solutions for scaling smart energy initiatives.

Future research

To enhance the generalizability and contextual robustness of the findings, future research should apply the AISILS framework across a broader set of geographic regions. Comparative case studies involving municipalities from other parts of the Western Balkans (e.g., Bosnia and Herzegovina, North Macedonia, and Montenegro), as well as from other emerging economies (e.g., the Caucasus, North Africa, and Central Asia), would enable researchers to test the framework’s adaptability under varied institutional, socio-technical, and climatic conditions. This would not only validate the AISILS model across different governance contexts but also uncover region-specific barriers and enablers of AI–solar integration. Cross-country comparisons would support the development of typologies of institutional readiness and inform regionally tailored policy recommendations for sustainable digital energy transitions.

The present study offers a foundational analysis of the synergistic role of solar power and artificial intelligence (AI) in promoting sustainable local development. To further consolidate and generalize the findings, several important avenues for future research are proposed. To understand the full sustainability potential of AI-enhanced solar deployments, future research should prioritize long-term, multi-year studies that capture the following:

• Seasonal Variability in PV Performance: Solar irradiance, energy generation patterns, and system degradation vary significantly across seasons. Longer observation windows (e.g., 24–36 months) would allow for more accurate assessment of solar performance under dynamic meteorological conditions and provide evidence for lifecycle planning.
• Lifecycle Emissions and Resource Use: Beyond operational efficiency, it is crucial to evaluate the embedded energy, carbon emissions, and material usage associated with PV panel production, AI hardware, and cloud infrastructure. This includes assessing cradle-to-grave emissions and estimating net environmental benefits of AI integration.
• Dynamic Socioeconomic Impacts: Economic and social effects—such as employment patterns, energy affordability, or gender inclusion—are likely to evolve over time. The longitudinal tracking of distributional outcomes and community wellbeing indicators is needed to assess whether AI–solar programs remain equitable and just.

While this study focused on four Serbian municipalities, comparative analyses across multiple countries in the Western Balkans and other emerging regions (e.g., Caucasus, North Africa, and Central Asia) can help:

• Refine and Validate the AISILS Framework: By applying the framework in diverse urban, peri-urban, and rural contexts with different institutional arrangements and socio-technical systems, researchers can evaluate its robustness, adaptability, and scalability.
• Assess Contextual Drivers and Barriers: Comparative studies can identify how regulatory environments, funding mechanisms, and cultural factors affect AI–solar integration. For instance, variations in digital literacy, political will, or citizen engagement may influence adoption trajectories.
• Develop Typologies of Institutional Readiness: Through clustering methods and diagnostic modeling, a typology of digital energy governance models across regions can be constructed, helping policymakers prioritize interventions based on maturity levels.

As AI systems become integral to energy planning and decision making, dedicated research must investigate the following:

• AI Ethics and Algorithmic Transparency: Questions around explainability, fairness, and potential bias in AI forecasting or resource allocation algorithms demand critical scrutiny. Studies should explore how these tools may inadvertently reinforce social inequities or obscure accountability in public sector energy decisions.
• Data Sovereignty and Governance: The deployment of AI relies on high-frequency, granular data, often stored in cloud infrastructures managed by third parties. There is a growing need for studies addressing ownership, control, localization, and regulatory oversight of municipal energy data in AI-based systems.
• Trust and Co-Creation Mechanisms: Public acceptance of digital energy transitions hinges on citizen involvement, participatory design, and transparent communication. Future research should explore how to embed community voices in the design, deployment, and evaluation of AI-enhanced systems, especially in vulnerable or underserved populations.

To guide evidence-based policy development, future studies should incorporate simulation and scenario modeling approaches, including techno-economic optimization models, to assess investment trade-offs and cost–benefit ratios for various AI–solar configurations under different policy incentives, tariff structures, and funding schemes. Also, it is critical to perform resilience and risk modeling, to evaluate the robustness of AI–solar power systems under extreme events (e.g., climate-induced grid stress and cyber threats), including the role of digital redundancy and decentralized intelligence. Finally, long-term sustainability project outcomes (e.g., emission reductions and SDG progress) must be tracked having in mind varying adoption rates of AI and solar integration across municipalities.