The growing environmental, social, and economic challenges of unchecked industrial development have led to the widespread adoption of sustainable development principles, rooted in balancing economic growth with environmental protection and social equity. Circular economy enterprises, which seek to minimize waste and optimize resource use, are increasingly recognized as key players in achieving sustainability. This research explores the potential of artificial intelligence (AI) to support circular enterprises by improving decision-making, enhancing resource efficiency, and reducing waste. Through AI-driven predictive analytics, automation, and data analysis, circular enterprises can optimize their processes, making strides towards sustainable operations while addressing challenges such as resource management, production efficiency, and supply chain optimization.
The study builds on foundational sustainable development theories while focusing on AI's role in advancing circular economy principles in small municipalities and small-to-medium enterprises (SMEs). In its initial phase, the research defines critical elements within circular enterprises, collects relevant data (including a nationwide survey of local governments), and identifies opportunities for AI integration to solve operational challenges. By aligning circular economy goals with AI capabilities, this work contributes an innovative framework that bridges the digital divide and drives businesses toward a more sustainable future.
Sustainable development emerged in the late twentieth century in response to the negative impacts of unchecked industrialization and economic growth. Foundational works such as the Club of Rome'sLimits to Growth (1972, with updates in 1992 and 2004) warned that continuing current growth trends could lead to catastrophic ecological and social (Meadows et al., 2004). The United Nations’ Brundtland Commission (1987) formally defined sustainable development as “development that meets the needs of the present without compromising the ability of future generations to meet their own needs”(World Commission on Environment and Developmen, 1987), emphasizing a balance between economic development, environmental protection, and social well-being (Barkemeyer et al., 2014; Imran et al., 2014). Over the ensuing decades, global initiatives like Agenda 21 from the 1992 Earth Summit and early reports by the Intergovernmental Panel on Climate Change (IPCC) reinforced the need for integrating these principles into policy and practice (United Nations, 1992). Scholars have noted that sustainable development remains an evolving ideal–widely endorsed yet difficult to achieve in practice (Holden et al., 2014), especially without the active involvement of businesses (Zgrzywa-Ziemak, 2019).
Within this sustainability context, the concept of the circular economy has gained prominence as a pathway for businesses to contribute to a sustainable future. Circular economy enterprises aim to design out waste and keep products and materials in use, representing an advanced form of “sustainable enterprise”. As Zgrzywa-Ziemak observes, truly sustainable enterprises are an ultimate goal rather than a present reality (Zgrzywa-Ziemak, 2019). Building on prior definitions of sustainable enterprises (e.g., Keijzers’ definition emphasizing reduced pollution, renewable energy use, and preservation of natural capital (Keijzers, 2002), circular enterprises implement business models that create value by minimizing waste and maximizing resource use throughout product lifecycles (Ellen MacArthur Foundation, 2013). According to the Ellen MacArthur Foundation, such enterprises design products for longevity, reuse, and recyclability (Ellen MacArthur Foundation, 2015), aligning with Kirchherr et al.'s vision of circular economy principles in business. Popular circular business models include:
Product-as-a-Service – offering products through rental or subscription instead of direct sale.
Extending Product Life – through maintenance, repair, and refurbishment to prolong usage.
Resource Recovery – reclaiming and recycling materials at a product's end-of-life for new uses.
Despite their promise, circular enterprises face significant challenges in practice. These include cost optimization and operational difficulties in resource management, production processes, and supply chains (Huang et al., 2024). Many firms struggle to adopt circular practices due to these barriers, particularly smaller organizations with limited capital or expertise. To overcome such challenges, businesses are increasingly exploring artificial intelligence (AI) as a tool for smarter decision-making and process optimization (Bashynska, 2025).
However, most prior research and implementations of AI in the circular economy context have focused on large corporations or advanced smart cities. This study uniquely bridges the application of AI to circular economy enterprises with a special focus on small municipalities and SMEs, which are often left behind in digital transformation. These smaller entities face a “digital divide” a gap in access to technology, expertise, and infrastructure–that hampers their ability to implement advanced AI-driven solutions. Previous works have not adequately addressed this gap, concentrating instead on well-resourced organizations. By contrast, our research targets the practical barriers faced by resource-constrained communities and proposes a scalable, low-cost AI-based prototype tailored to their needs. In doing so, we emphasize the novelty of our approach: applying generative AI techniques in a cost-effective manner to support circular initiatives at the local level, thereby addressing technological and financial limitations not previously discussed in circular economy research.
The main goal of this research is to identify and analyze the challenges that circular enterprises face in their operations, particularly in small-city contexts, and to develop methods to improve resource efficiency and reduce waste through innovative, AI-driven approaches. We posit the following hypothesis: AI-driven approaches have significant potential to enhance the performance of circular enterprises, but their effective implementation is hampered by the digital divide, resulting in uneven adoption and outcomes across communities. To investigate this hypothesis, our study combines theoretical and empirical analysis, including a literature review, a bibliometric study of research trends, and a nationwide survey of local government units in Poland.
Urban sustainability challenges provide a pertinent context for this research. Notably, the United Nations’ Sustainable Development Goal 11 (“Sustainable Cities and Communities”) highlights several pressing issues that city administrations–especially in developing and transition economies–must address to become inclusive, resilient, and sustainable (United Nations, 2015). Key challenges include:
Rapid urbanization – leading to overpopulation, strained infrastructure, and informal settlements lacking basic services.
Environmental degradation – pollution of air, water, and soil, loss of green spaces, and reduced biodiversity affecting quality of life.
Aging and insufficient infrastructure – outdated transport, water, and energy systems, and inadequate waste management and sanitation services.
Social inequalities – unequal access to education, healthcare, and economic opportunities, with marginalized groups often lacking essential services.
Climate vulnerability – increasing exposure to floods, heatwaves, and extreme weather, requiring adaptation and resilience measures.
Transportation and mobility issues – traffic congestion, high emissions, and limited public transit or non-motorized transport options.
Small and mid-sized municipalities frequently struggle with these issues under tight budget constraints and limited technical know-how. This context underscores the importance of smart, resource-efficient solutions–precisely the kind that AI can enable–to help cities and enterprises transition toward circular, sustainable operations. Our research is positioned at the intersection of these trends, aiming to contribute a framework that leverages AI to advance circular economy goals while explicitly bridging the digital divide that affects less advantaged communities.
Literature Review
Sustainable Development and Circular Economy
Achieving sustainability in business operations has been a subject of extensive study. No enterprise today can be deemed fully “sustainable” in practice–it is an aspirational goal requiring continuous improvement (Zgrzywa-Ziemak, 2019). As such, researchers have developed frameworks and definitions to guide businesses toward sustainability. Keijzers (2002), for example, described a sustainable enterprise as one whose processes reduce pollution and waste, utilize renewable resources, preserve essential natural capital, and simultaneously enable social and economic development (Keijzers, 2002). This definition highlights that sustainability extends beyond environmental management to include economic viability and social responsibility (Ganushchak, 2025).
Within this paradigm, the circular economy has emerged as a practical approach to operationalize sustainability. The circular economy concept moves away from the traditional linear “take-make-dispose” model towards a regenerative system of production and consumption. Kirchherr et al. (2017) identified over 100 definitions of the circular economy, generally centering on resource loops and minimizing waste (Kirchherr et al., 2017). The Ellen MacArthur Foundation defines a circular enterprise as one that implements circular principles in its business model–creating value by keeping products and materials in use and designing out waste (Ellen MacArthur Foundation, 2013). Key strategies include designing products for durability and reuse, and establishing systems for recycling and recovering materials. In essence, a circular enterprise strives to treat waste as a resource and to maintain the value of products and materials for as long as possible.
Research has catalogued several business models that enable circular operations, as noted above: product-as-a-service, extending product life, and resource recovery. Each model requires changes in how firms create and deliver value. For instance, offering a product as a service (e.g., leasing machinery instead of selling it) incentivizes the provider to maximize the product's lifespan and maintainability. Likewise, focusing on refurbishment and recycling can cut costs and reduce environmental impact. These models have been successfully demonstrated by large companies (for example, Caterpillar's engine remanufacturing program, discussed later) and are increasingly being considered by smaller firms and startups.
Despite clear environmental and economic incentives, the transition to circular business models faces significant barriers. Companies often encounter high upfront costs for new technologies, difficulty in redesigning supply chains, regulatory hurdles, and uncertainty about long-term benefits. Moreover, the expertise needed to implement circular practices (such as lifecycle assessment, reverse logistics, or new materials science) may be lacking, especially in SMEs. This is where digital technologies, and AI in particular, are seen as powerful enablers to help overcome complexity and optimize circular processes. Before exploring AI's role, it is instructive to review recent studies that have examined how enterprises pursue circularity.
Recent literature (2022–2024) has increasingly focused on integrating advanced technologies into circular economy initiatives. Huang et al. (2024) present a four-stage circular enterprise maturity model, illustrating how companies can progressively enhance their circular practices with matching AI techniques at each stage. This model provides practical guidance for managers to assess their current level and identify suitable AI tools as they advance (e.g., from basic recycling efforts to fully data-driven circular supply chains). Bashynska (2025) explore AI applications in circular economy contexts through case studies, demonstrating how AI-driven analytics can improve resource optimization, sustainable supply chain management, and circular-compliant manufacturing (Bashynska, 2025). Their findings show tangible efficiency gains and waste reductions when AI is applied to monitor production processes and materials flows. In the domain of municipal waste management, Czemiel-Grzybowska et al. (2024) describe the use of AI to optimize operations in waste incineration plants, illustrating how smart technologies can enhance technological entrepreneurship in smart city initiatives (Czemiel-Grzybowska et al., 2024). Using a mix of case study analysis and expert interviews, they found that AI-based systems (e.g., for combustion control or maintenance scheduling) significantly improved plant performance and aligned with sustainable development goals (Sachs, 2015). These recent works collectively indicate a growing recognition that AI can be a catalyst for circular economy adoption, by handling complex data and improving decision quality in areas ranging from manufacturing to municipal services (Bashynska & Dyskina, 2018).
Earlier foundational studies also support the synergy between digital technology and circular strategy. Kristoffersen (2020) introduced a digital-enabled circular strategies framework for manufacturing companies, highlighting how data analytics and connectivity can boost circular outcomes in production environments (Kristoffersen, 2020). Nižetić (2019) discussed smart technologies for energy efficiency and waste management, suggesting that IoT sensors and AI algorithms can significantly improve resource utilization and waste reduction in industrial (Nižetić, 2019). Similarly, Li et al. (2021) proposed a data-driven reversible framework for sustainable product-service systems, which uses real-time data and AI to dynamically adjust services for sustainability gains (Li et al., 2021). These studies, while not focused on small organizations, provide useful concepts and demonstrate the versatility of AI in advancing circular principles.
In summary, the literature establishes that integrating AI and related digital tools can greatly assist circular enterprises. At the same time, it reveals a gap: most research and examples pertain to relatively large or technologically advanced entities. There is a notable lack of studies addressing small municipalities or SMEs and how they can leverage AI for circular economy initiatives (Czemiel-Grzybowska et al., 2024). This gap has been quantitatively confirmed by our bibliometric analysis (detailed below), which found very few publications on “smart circular enterprises” in the context of smaller communities. Our work aims to fill this gap by focusing on bridging the digital divide–ensuring that AI's benefits for circular economy are accessible to less-resourced organizations.
Role of AI in Circular Economy
AI technologies offer powerful tools to improve the efficiency and effectiveness of circular economy processes (Lewicka et al., 2025). A review of the literature identifies several key contributions of AI to circular enterprises:
Predictive Analytics. AI can forecast resource needs, demand patterns, and potential shortages. For example, machine learning models can predict when certain materials will run low or when a spike in waste generation might occur, allowing businesses or city managers to prepare and allocate resources optimally. Predictive maintenance algorithms can also anticipate equipment failures, reducing downtime and extending the life of machinery in production (LinkedIn, 2025; Pathan et al., 2023).
Automation. AI-driven automation can streamline production and recycling processes. Robotics and intelligent control systems can sort waste materials, disassemble products for remanufacturing, or adjust manufacturing parameters in real time to reduce material waste and energy consumption (Moloney & Raad, 2025; Holzhausen,2024). Automation not only cuts labor costs but also improves consistency in processes like material recovery or precision manufacturing, thus supporting circularity.
Data Analysis & Optimization: Circular enterprises generate and rely on large amounts of data – from supply chain data to sensor readings in recycling facilities. AI excels at processing big data to uncover inefficiencies and optimization. For instance, AI can analyze supply chain logistics to minimize transport distances (reducing emissions) (Choubey & Karmakar, 2021; Sneed, 2017) or identify patterns in consumer returns that could inform better product design (Ellen MacArthur Foundation, 2021). Complex trade-offs (such as choosing materials that are both low-impact and high-performance) can be evaluated using AI optimization models (Kuang et al., 2021).
Supply Chain Coordination. In a circular economy, supply chains often become supply loops, with reverse logistics (returns, recycling) as important as forward logistics. AI can greatly enhance supply chain management by predicting demand, optimizing inventory levels, and dynamically routing shipments (Prism Sustainability Directory, 2025). This is crucial for circular systems where timing and coordination (e.g., collecting used products for refurbishment) affect viability. AI-driven platforms can match waste outputs of one process as input materials for another, facilitating industrial symbiosis.
In broad terms, artificial intelligence refers to computer systems capable of performing tasks that typically require human intelligence, such as learning, reasoning, pattern recognition, and decision-making (Ellen MacArthur Foundation, 2021). Modern AI systems often employ machine learning and neural networks, which learn from historical data to improve their performance without explicit programming for each scenario (Choubey & Karmakar, 2021). These systems can adapt to new inputs and provide insights or predictions that would be difficult to derive with manual analysis. The concept of AI has evolved since the seminal Dartmouth Conference in 1956 (where McCarthy et al., framed the goal of making machines behave intelligently) and today encompasses a range of techniques from rule-based expert systems to advanced deep learning models (McCarthy, 1955). A subset of AI particularly relevant to our work is generative AI, which can create new content or solutions (e.g., generating optimized plans or simulating scenarios) based on learned patterns.
AI's relevance to the circular economy has been explicitly recognized in recent strategic reports. For example, the Ellen MacArthur Foundation (2021) discussed AI as a tool to accelerate the transition to a circular economy, noting its potential in pattern recognition, prediction, and optimization tasks that are essential for managing complex circular systems (Ellen MacArthur Foundation, 2021). In practical terms, many industries are already leveraging AI to drive sustainability: energy companies use AI to balance renewable power grids, manufacturers use AI for quality control to reduce waste, and cities use AI to optimize traffic flows and reduce emissions. These applications align with circular economy goals by increasing efficiency (doing more with fewer resources) and preventing waste and pollution.
By integrating AI solutions, circular economy enterprises can enhance their sustainability performance while remaining competitive (Ochieng et al., 2024). AI enables a shift from reactive to proactive management–problems can be anticipated and prevented rather than simply dealt with after the fact. For instance, instead of reacting to landfill overflows, a city could use AI to forecast waste generation and expand recycling capacity in advance. Instead of dealing with resource shortages, a company could use AI to design products that require less of a scarce material or to find secondary materials before shortages occur. In effect, AI helps enterprises move toward zero-waste operations and maximal resource productivity, aligning business success with environmental stewardship (Nižetić, 2019; Kristoffersen, 2020).
In conclusion, the literature indicates that companies adopting circular economy models should prioritize long-term resource management, extend product life cycles, and minimize waste–and that AI will be a vital tool in achieving these aims (Li et al., 2021). Particularly for the challenges identified (complex supply chains, need for efficiency, data-heavy decision processes), AI offers capabilities that traditional methods cannot easily provide. This sets the stage for our research, which seeks to apply these insights in a novel context (small municipalities and SMEs) and demonstrate how AI can be harnessed to overcome the digital divide in circular economy adoption.
Research Methodology
To test our hypothesis and achieve the research objectives, we adopted a multi-faceted methodology with both qualitative and quantitative components. The overall research design is illustrated in a conceptual framework (Figure 1) comprising sequential stages that examine the interplay between sustainable development concepts, circular enterprise challenges, AI as an enabler, and the digital divide, all within the context of municipal enterprises in Poland.
Conceptual Framework of the Study.
In summary, the research proceeded through the following steps:
Foundational concepts. We began by reviewing the established principles of sustainable development and circular economy. This provided the theoretical grounding and key definitions for our study.
Challenges for circular enterprises. Next, we identified specific challenges that circular economy enterprises face, particularly focusing on resource management, production, and supply chain issues. We postulated that these operational challenges can be exacerbated by a digital divide–i.e., uneven access to advanced technologies and know-how–leading to limited implementation of circular solutions in less technologically advanced communities.
AI as an enabler. We then explored the potential of AI-driven solutions to address the above challenges. We theorized how AI could improve efficiency and effectiveness of circular operations (e.g., by optimizing resource use and waste reduction), and thereby help bridge gaps caused by the digital divide. This stage informed the design of our proposed AI framework.
Empirical research methodologies. Two primary research methods were employed to evaluate our hypothesis and inform the framework: a bibliometric analysis and a survey of local government units. The bibliometric analysis identifies existing research trends and gaps in the literature, while the survey gathers real-world data on challenges and current practices (Herlina et al., 2024; Merton, 1967). These methods are complementary – one mapping the knowledge landscape, the other providing on-the-ground insights (Bemke-Świtilnik, 2018).
Integration and solution development. Finally, we integrate findings from the literature, bibliometric analysis, and survey to develop a prototype AI-based solution. This involves formulating a framework (and a conceptual system architecture) for a generative AI tool tailored to small municipalities, designed to be low-cost and scalable via cloud technologies. The proposed solution directly targets the barriers identified, with the ultimate goal of enhancing circular economy practices in the local context.
Following this research model, the bibliometric analysis was conducted first to gain a broad overview of relevant research and to substantiate the novelty of our focus; to explore research trends within specific areas of interest, namely smart circular enterprises, AI-driven approaches, and issues related to the digital divide. The analysis used data from the Scopus SciVal database, a recognized and scientifically credible platform by Elsevier, focusing on articles published from 2019–2024 and utilizing three meticulously selected keywords to ensure the relevance of the findings. The analysis yielded a total of 3,256 articles and 3,399 citations, extracted from the bibliographies of prominent Scopus-indexed journals during the specified timeframe.
We gathered publication data from 2019–2024 using key terms related to our topic (e.g., “AI-driven approaches”, “digital divide”, “smart circular enterprises”). Using academic databases, we identified the top journals publishing work on these keywords and counted the number of articles per year. This analysis revealed some striking trends. For instance, “AI-driven approaches” in sustainability saw a surge of interest in that period (roughly a 1000% increase in related publications), and “digital divide” topics also rose significantly (∼145% increase). This phenomenon is illustrated in the graph below Figure 2.
Published Articles (2019–2024).
Based on this upward trend, it is expected that the number of scientific publications on the digital divide will continue to grow over the next 3–4 years. By 2027, the volume of articles may exceed 1,000 annually, driven by the increasing global focus on digital equity, AI accessibility, and smart infrastructure development. In contrast, “smart circular enterprises” appeared in far fewer publications–peaking around 2021 and then declining. In total, we found only 7 articles explicitly discussing “smart circular enterprises” by 2024, compared to over 3,200 articles on the digital divide. This disparity underscores a gap in the research: while digitalization and AI in sustainability are well-studied, their application in circular economy at the enterprise level (especially for smaller communities) is underrepresented. Our work is positioned to help fill this gap by focusing on that niche.
The second phase of our methodology involved an empirical survey to collect data on current practices, challenges, and perceptions regarding circular economy and AI at the local government level. We designed a structured questionnaire comprising three key questions and distributed it to all local government units across Poland. In the Polish context, these units (gminas, poviats, and voivodeships) are the administrative bodies responsible for city management and local economic development, making them ideal respondents for understanding community-level circular initiatives. We compiled a contact database from open public sources (with the assistance of Poland's Act on Access to Public Information) and sent out the survey in fall 2024 (September-November). In total, 2,831 official units were contacted with the questionnaire (Baza JST, 2024). The survey questions were as follows:
“Are there any circular (closed-loop) enterprises or initiatives in your jurisdiction as part of public city management? If so, what kind of enterprises or what profile do these initiatives have?” – (This aimed to gauge awareness and existence of circular economy practices at the local level.)
“Are artificial intelligence (AI) technologies used in city management or in enterprises supporting your city's functions? If yes, in what areas are these AI solutions applied?” – (This addressed the digital aspect, identifying any current usage of AI or smart technologies in municipal services or local businesses.)
“What are the biggest challenges that your city/local government unit currently faces in terms of sustainable development and meeting the conditions of circular solutions?” – (This invited open-ended responses about barriers, which could include financial, technical, social, or regulatory challenges.)
The survey primarily used multiple-choice and short-answer formats to facilitate a high response rate and easy analysis. Out of the 2,831 units contacted, we received 521 valid responses, which is an 18.4% response rate. The respondents included officials from various levels of local administration, ensuring a broad representation of municipalities by size and region. We assessed the quality and clarity of the responses and found them generally informative for our research purposes Figures 3 and 4.
Correlation of Units That Took Part in the Survey.
Correlation of Applicable Survey Responses.
For data analysis, we first employed descriptive statistics to summarize the survey results and identify prevalent themes. This involved calculating response frequencies and percentages for each question and grouping common answers (especially for the open-ended third question). We also categorized the responding units by population size: we defined “small” municipalities as those with <20,000 inhabitants, “medium” as 20,000–100,000, and “large” as >100,000. This clustering allowed us to observe differences in challenges and technology adoption across different scales of communities. Indeed, preliminary analysis suggested notable differences–for example, certain challenges were reported far more frequently by small towns than by larger cities. These insights guided our development of the AI framework to ensure it addresses the most pressing needs of smaller units.
We recognize that a deeper statistical analysis is needed to draw robust conclusions about relationships in the data (for instance, the correlation between municipality size and likelihood of using AI, or between budget levels and types of challenges reported). Therefore, future work will incorporate inferential statistical techniques. In particular, we plan to perform Pearson correlation analysis to quantify the association between municipality characteristics (like size or budget per capita) and specific reported barriers, and Chi-square tests of independence to determine if differences in responses (e.g., the proportion citing a given challenge) are statistically significant across groups (Regiony Polski, 2024). These tests will help validate whether the trends observed (e.g., smaller municipalities more often citing lack of funds) are unlikely to be due to chance. For the scope of this paper, however, we limit the discussion to descriptive findings, which already provide valuable guidance for the framework design.
In summary, our methodology combines a literature/bibliometric review to establish context and novelty, with a comprehensive survey yielding empirical data on current circular economy and AI practices in Polish municipalities. This mixed-methods approach ensures that our proposed solution is grounded in both the state-of-the-art research and real-world conditions. The next sections present the results of the survey and the insights gained, followed by the proposed AI-driven framework and its evaluation in a broader context.
Results
Survey Results
The survey of Polish local government units (LGUs) offers a snapshot of how circular economy concepts and AI technologies are (or are not) being adopted at the community level, as well as the obstacles faced. Out of 2,831 LGUs contacted, 521 responded (approximately 18.4%). These included a mix of urban municipalities and rural communes, providing a diverse perspective. We organized the responses by municipality size to better interpret the data:
Small municipalities (<20k population): 350 responses (67% of total)
Medium municipalities (20k–100k): 140 responses (27%)
Large municipalities (>100k): 31 responses (6%)
(This breakdown roughly reflects the national distribution of municipality sizes, skewed toward many small communities and fewer large cities). A detailed breakdown of responses by clusters and by territory is presented in Figures 5 and 6.
Responses Division According to Clusters.
Responses Division According to Territorial Spread.
In future research stages, inferential statistical techniques such as Pearson correlation and Chi-square independence testing will be employed to assess the significance of observed relationships between municipality characteristics and AI adoption barriers.
Circular economy initiatives (Q1). A majority of respondents acknowledged at least some ongoing initiatives related to circular economy or sustainability in their locality. Many mentioned traditional recycling programs or waste segregation efforts. A few cited more advanced projects, such as bio-composting facilities, water reuse systems, or partnerships with private “green” enterprises (like companies that use municipal waste as raw material). However, in smaller towns, 65% indicated “no significant closed-loop enterprises” in operation–highlighting that circular economy concepts are often still in nascent stages outside major cities. Larger cities were more likely to report specific programs (e.g., a circular food economy project or an industrial symbiosis park in their region). Overall, the data suggest that awareness of circular economy exists broadly, but concrete implementations are uneven and often limited in scope.
Use of AI technologies (Q2). The adoption of AI or smart technologies in local governance or local enterprises appears quite limited, especially among smaller municipalities. Around 80% of small municipalities reported that they are not currently using any AI-based solutions in city management. Some of the remaining 20% referenced basic “smart city” tools (for example, intelligent street lighting systems or simple sensor-based monitoring of air quality), rather than AI per se. Medium and large cities showed somewhat higher adoption–about half of the large municipalities reported using AI or advanced data analytics in areas such as traffic management, public safety (e.g., smart CCTV analytics), or optimization of public transport routes. A notable example from a medium-sized city was the use of a machine learning system to predict water usage patterns for better reservoir management. Still, the overall picture is that AI usage in municipal operations is low, and practically nonexistent in smaller communities. This underscores a significant digital divide: while leading global cities experiment with AI for various services, most Polish localities have yet to implement even entry-level AI solutions.
Key challenges (Q3). The open-ended question about challenges elicited detailed responses. Several common themes emerged, which we group into three broad categories: financial constraints, low public awareness, and regulatory hurdles and governance issues. The most frequently cited challenges by local authorities include:
Financial constraints. Nearly every respondent (especially from small and medium municipalities) mentioned lack of funds as a critical barrier to sustainable development initiatives. Limited budgets make it difficult to invest in modern technologies or infrastructure upgrades needed for effective waste management, renewable energy, or other circular solutions. For example, establishing comprehensive recycling systems or circular waste processing facilities requires significant capital and long-term planning, which is hard to justify under tight fiscal conditions. As one respondent noted, “Creating systems for segregation, recycling, and reuse requires significant investment and long-term planning, which is difficult given our limited city budget.” This challenge was identified as the primary obstacle by over 70% of small municipalities. Larger cities also acknowledge financial issues, but they often have relatively more resources or access to external funding (EU funds, national programs) to mitigate this issue.
Low public awareness and engagement. The second major theme is the lack of community awareness or buy-in regarding sustainability and circular practices. Many local officials observed that residents often do not understand the benefits of recycling, energy saving, or other ecological behaviors, which hampers the adoption of circular initiatives. Habit change is slow; for instance, attempts to introduce composting or reduce single-use plastics met with low participation in some towns. “Residents may not be aware of the benefits of more ecological practices, making behavior change a long-term process,” one response noted. Education campaigns and public outreach are needed but can be time-consuming and require effort from the already resource-strapped municipal authorities. Larger cities have typically run environmental awareness programs for years (often with support of NGOs or national campaigns), resulting in relatively better public awareness – which explains why this challenge was emphasized far more by small municipalities than big cities. In smaller towns, officials feel that sustainability has not been a priority historically, so public knowledge starts from a low base.
Regulatory and institutional hurdles. A third common challenge is navigating the complex and evolving legislative environment around waste management, recycling, and environmental protection. Respondents pointed out that laws and regulations change frequently, and keeping up (and complying) is difficult, especially without dedicated legal or environmental staff. Bureaucratic complexity in obtaining permits or approvals for new projects (e.g., a waste-to-energy plant or even installing solar panels on public buildings) was cited as a deterrent. Some mentioned the lack of clear national guidelines or the instability of regulations with changing governments, which makes long-term planning for circular economy projects risky (Filyppova et al., 2019). Additionally, successful circular economy efforts often require coordination across sectors (public, private, NGOs) and across municipal boundaries – something that is challenging to organize. Smaller municipalities in particular noted they lack leverage and expertise to form effective partnerships for circular initiatives. As one response summarized, “We must implement new national and EU regulations, which often come with additional costs. The instability of these regulations and frequent changes make it difficult to plan long-term sustainable development strategies.” On the positive side, a few larger cities mentioned that they are forming multi-sector collaborations (with businesses and neighboring municipalities) to tackle issues like air quality and waste, suggesting that with more capacity, these challenges can be partially overcome.
In addition to the above, the survey responses shed light on practical problem areas that cities are currently grappling with on their path to sustainability and circularity. The most commonly mentioned areas of concern were:
Waste management. Almost all cities, regardless of size, highlighted the issue of effective waste management and the need to reduce waste sent to landfills. Growing populations and consumption have led to increased waste generation, and many cities struggle to expand or improve their recycling systems. Meeting higher recycling targets and reducing landfill dependency is seen as essential but difficult without new investments or technologies.
Water and wastewater management. Ensuring access to clean water and treating sewage effectively was another significant challenge. Pollution of water bodies and the need for advanced wastewater treatment (possibly for reuse) were mentioned, especially in context of climate change causing water stress.
Sustainable transport. Many urban respondents stressed the challenge of reducing traffic congestion and emissions. Solutions suggested include investing in public transport, cycling infrastructure, and transitioning city fleets to electric vehicles – all requiring funding and planning.
Preservation of green spaces. Maintaining and expanding green areas (parks, urban forests) is important for quality of life and biodiversity, yet cities face pressure from development and budget constraints in managing these spaces. Respondents acknowledge the value of urban greenery for microclimate regulation (mitigating heat islands, etc.) and recreation, but need support to protect these areas.
Despite limited resources, local governments are undertaking small-scale actions to address some of these issues. Common immediate measures include: energy retrofits of public buildings (improving insulation, upgrading heating systems to more efficient or renewable-based ones), establishing selective waste collection points (to facilitate recycling of different waste streams), and running educational campaigns in schools and communities about waste segregation and recycling. Some cities have introduced subsidy programs for residents to replace old polluting home heating systems (“smokestack” boilers) with cleaner alternatives, as a way to improve air quality and progress toward circular energy use. These measures are often cited as the “simplest minimal actions” that can be done under current financial limitations. They have yielded incremental benefits, but officials recognize that bolder initiatives will require much greater support and capacity.
Identified Barriers and Opportunities
Analyzing the survey findings in depth reveals underlying structural barriers that particularly hinder smaller municipalities, as well as potential opportunities to support them. A clear disparity emerges between large, well-resourced cities and smaller, resource-constrained towns. Larger cities (often with diversified economies and bigger tax bases) reported relatively fewer barriers in funding and capacity–some have been able to invest in renewable energy projects, modernize infrastructure, and implement sustainability programs. They often benefit from economies of scale and can attract external funding or private partners. In contrast, over 70% of small municipalities identified financial constraints as their primary challenge to implementing circular solutions. These towns frequently lack major industries or “anchor” enterprises that could drive or fund circular initiatives, leading to a vicious cycle of limited local revenue and low investment in innovation.
Several respondents from small towns pointed out consequential issues stemming from this situation. When a municipality cannot progress on sustainability or offer modern amenities, it tends to suffer from out-migration of young talent (as educated youth move to larger cities with better opportunities). This brain drain further reduces the local capacity to implement new technologies (including AI)–there may simply be no skilled personnel to lead such projects. Additionally, a lack of visible development can deter business investment, trapping the town in an economic stagnation. In our survey, the most commonly cited reason for “lack of action in introducing sustainable development and circular economy principles” was precisely this lack of financial resources and investment. Officials described a scenario where without external support, they cannot afford projects that would make the locality more sustainable and attractive, and without those improvements, they cannot attract the investments that might improve their finances–a classic vicious cycle.
This analysis underscores a crucial point: the digital divide and what might be termed a “sustainability divide” go hand in hand. Smaller municipalities not only lag in digital infrastructure (broadband access, IT systems, etc.) but also in human capital and institutional capacity. The survey responses explicitly noted that many such towns have increasing “technological debt”–falling further behind each year as technology advances elsewhere. The consequences are stark: some officials fear long-term depopulation and economic decline if they cannot modernize. Indeed, many small towns indicated they are becoming disproportionately dependent on state aid and social welfare, rather than contributing to economic growth. This trend is unsustainable for the nation as a whole, highlighting that helping these communities adopt new solutions (like AI for circular economy) is not just a local issue but a national priority.
On the other hand, the results also highlight opportunities and potential leverage points. Larger cities in Poland have shown that with proper investment and planning, progress is possible–for example, some have successfully implemented solar farms, modern waste processing facilities, or comprehensive bike-sharing programs. These could serve as models or pilots that, if costs can be reduced, might be transferred to smaller towns. Moreover, Poland can learn from international examples (as we will discuss in the next section) where even resource-limited regions have leapfrogged by adopting innovative approaches and partnerships.
The survey indicates strong interest from local officials in solutions that are low-cost, scalable, and easy to implement given limited expertise. Many respondents welcomed the idea of centralized support or shared services–for instance, a cloud-based system provided at the national or regional level that municipalities could use for data analysis or planning. This kind of approach could help bypass the need for each small town to hire its own AI specialists or build expensive infrastructure. There is also openness to training programs that build digital skills among existing staff, as well as to collaborations (e.g., several neighboring towns pooling resources to hire a sustainability expert or to jointly invest in a recycling facility).
In summary, the empirical analysis paints a picture of a two-speed scenario: larger Polish cities gradually advancing in sustainability (and even beginning to use digital tools), versus small municipalities struggling with basic issues. Bridging this gap will require targeted interventions. The findings reinforce our hypothesis–AI has great potential to enhance circular enterprises, but the digital divide is a real and pressing barrier. Recognizing this, our study moves on to propose an AI-driven framework specifically tailored to the needs and constraints of small municipalities, in hopes of turning some of these barriers into opportunities for progress.
Discussion
Proposed AI-Driven Framework for Circular Enterprises
In response to the challenges identified through the survey and analysis, we propose a conceptual AI-driven decision support framework designed for small municipalities and local circular enterprises. The framework's primary objective is to help local decision-makers optimize resource use and waste management despite limited expertise and budgets. By doing so, it aims to directly address the financial and technological barriers (lack of funds, lack of skilled personnel, low awareness) that were highlighted by our respondents.
To complement the proposed research model with a technical layer, Figure 7 illustrates the conceptual architecture of the AI prototype developed for use in small municipalities aiming to implement circular economy strategies. The diagram presents the logical flow of data and analytical components that comprise the system. The process begins with the collection of input data from local sources, including variables such as waste volume, energy consumption, municipal budget, and population size. These inputs are processed through two core analytical modules. The first module, Clustering (K-Means), segments operational data into meaningful clusters, identifying patterns such as neighborhoods with high energy consumption or inefficient waste management practices. The second module, Forecasting (ARIMA or RNN), is designed to predict resource demands and potential critical events, such as seasonal waste surges or budget overruns. Both modules feed into an Insights Engine, which synthesizes results and generates tailored recommendations for local decision-makers. The final output is a low-cost smart initiative, adapted to the financial and technical realities of smaller communities. This architecture not only supports data-informed decision-making but also addresses digital divide challenges by providing scalable, cloud-compatible solutions aligned with sustainability goals (SDG 11). As such, the AI prototype serves as a practical enabler for enhancing resource efficiency, waste reduction, and strategic planning within circular enterprises at the local level.
Architecture of the AI Prototype for Smart Circular Enterprises.
The prototype framework was modeled with an archetypal small town in mind–one that struggles with budget constraints, has few or no dedicated sustainability staff, and faces low public awareness of circular practices. Notably, one of the surprising findings was the extent of the awareness problem: many small-town officials noted that residents’ lack of understanding of sustainability initiatives is a major hurdle, whereas in large cities public engagement is higher due to long-term educational efforts. Therefore, a component of our framework is to provide easily interpretable insights and recommendations that can also be communicated to the public to raise awareness (for example, identifying inefficiencies and showing how addressing them saves money and environment).
The proposed solution adopts a generative AI approach powered by real municipal data. The idea is to leverage the computational power and sophisticated algorithms made available by modern AI–often through cloud services–to perform analysis and forecasting that local human staff would find very difficult or time-consuming. Importantly, by using cloud-based AI platforms, we can make high-end computational resources available to small communities at a fraction of the cost (since they do not need to own the hardware or develop the software from scratch). This effectively turns the digital divide into a bridge: small municipalities can tap into “AI as a service” offerings provided by global tech companies or national providers. As long as they can upload their data (even minimal datasets) and interpret the outputs, they gain access to capabilities that individually they could never afford. This approach aligns with the notion that many AI tools today (like machine learning libraries, cloud AI APIs, etc.) are commoditized and can be used by non-experts via user-friendly interfaces or with minimal training.
In our framework, the role of AI is primarily to analyze large amounts of diverse, raw data (which would be difficult for a human to process) and generate actionable insights. For instance, by feeding in data on monthly waste volumes, energy usage, budget allocations, and service levels, the AI can detect patterns and inefficiencies–perhaps identifying that certain neighborhoods consistently underperform in recycling, or predicting that waste collection can be rescheduled to be more efficient. The AI can also forecast potential critical periods (e.g., a likely surge in waste after holiday seasons or an impending budget shortfall in winter due to heating subsidies). By foreseeing these issues, local authorities can take proactive measures (like arranging extra recycling drives or adjusting budget plans) rather than reacting late when problems become acute.
It must be emphasized that the AI framework is a decision-support tool, not an autonomous decision-maker. Final decisions will and should be made by human officials, who can factor in context, values, and political considerations. The AI serves to reduce uncertainty and provide evidence-based suggestions, thereby lowering the risk and cost of decisions. As one of our principles, decisions supported by data (and AI analysis) are likely to be less risky, less costly, and more proactive, because they allow officials to anticipate issues and address root causes rather than just symptoms.
Framework Components and Algorithms
The AI-driven framework consists of several modules working in sequence to turn data into recommendations. It leverages both classical statistical models and modern machine learning algorithms to cover different needs:
Data ingestion and processing. The system first collects relevant data from the municipality. This may include data on waste collection (tonnage, frequency), energy consumption in public facilities, water usage, population statistics, budget figures, etc. Data quality is crucial – existing databases (if any) can be connected, and if data is missing, part of the effort would involve setting up simple data collection (even manual inputs) for key variables. The framework is flexible in accepting whatever data is available; if a town only has basic data (like monthly waste totals), the AI will work with that and perhaps recommend additional data to collect over time. Data is cleaned and normalized in this step.
Segmentation (clustering) module. We incorporate a clustering algorithm (such as K-Means clustering) to segment the operational data into meaningful groups. For example, the algorithm might cluster neighborhoods or time periods with similar characteristics. In a waste management context, clustering could reveal groups of neighborhoods – say, one cluster of areas with high waste generation and low recycling rates, and another cluster with opposite characteristics. By identifying these clusters, officials can tailor interventions (focus education campaigns on the cluster that performs poorly, or investigate why another cluster does well and replicate those conditions). Clustering can also be applied to budget data or other metrics to find which areas of spending or operations are outliers. The strength of an unsupervised learning approach like clustering is that it can highlight patterns without prior assumptions.
Forecasting module. To address the need for prediction, we include a time-series forecasting component. This can use models like ARIMA (Auto-Regressive Integrated Moving Average) for simpler forecasting tasks or more advanced Recurrent Neural Networks (RNNs) for complex patterns, depending on data availability. Forecasts are generated for key variables such as waste volume trends, energy demand, or even population changes. For example, using a few years of monthly waste data, the model can forecast waste generation for the next year, identifying peak months that might strain capacity. ARIMA is suitable for relatively short-term, linear trends, while an RNN (or its variant LSTM) could capture seasonality and non-linear patterns if a longer historical dataset is available. In many small municipalities, data may be sparse, so initially ARIMA might be used (needing less data), and as more data is collected or if regional data is pooled, an RNN could be explored. The forecasting module helps anticipate “potential critical events” such as seasonal surges in waste or budget overruns given current spending trajectories.
Insights engine. The outputs of clustering and forecasting feed into an Insights Engine. This engine synthesizes the findings and generates human-readable insights and recommendations. For instance, it might output: “Neighborhoods A, B, C produce 40% more waste per capita than the town average – consider increasing recycling facilities or education in these areas,” or “Waste volumes are projected to increase 15% in December – plan for additional collection rounds or temporary storage”. The insights engine can also integrate domain knowledge (rules-of-thumb, best practices) with the AI results. It essentially forms the “expert system” layer that translates data science results into specific actions or policy suggestions.
User interface and reporting. The framework would ideally have a simple dashboard interface for municipal officials to view the findings. It could display trends (graphs of forecasts), maps highlighting clusters or hotspots, and a list of recommendations ranked by priority or impact. This interface plays a role in bridging the gap with human users, ensuring the AI's output is transparent and understandable. For example, if the AI flags an anomaly, the official should be able to click and see the underlying data trend that led to that flag. This transparency will help build trust in the system and also serve educational purposes (showing officials the value of data-driven analysis, potentially improving their own analytical skills over time).
In terms of Key Performance Indicators (KPIs) for evaluating the success of implementing this framework, we propose the following metrics, which align with both circular economy objectives and the specific needs identified (Table 1).
KPIs for Evaluating the Effectiveness of the AI-Driven Circular Economy Framework.
KPI
Description
Example Metrics
Resource usage reduction
Reduction in resource consumption (e.g., electricity, water) achieved through AI-driven recommendations.
% decrease in energy or water consumption by municipal operations year-over-year.
Cost savings
Reduction in operational costs or avoidance of unnecessary expenses through process optimization, such as route and schedule adjustments.
% reduction in waste management costs; fuel and labor cost savings.
Waste reduction & diversion
Decrease in waste sent to landfills and/or increase in recycling and reuse rates, directly reflecting circular economy impact.
% increase in recycling rate (e.g., from 20% to 30% over two years); reduction in landfill waste per capita.
Service efficiency improvement
Improvement in service reliability and response times by predicting failures and proactively addressing issues.
% reduction in equipment downtime; shorter response time to citizen complaints.
Source: Prepared by the Authors (2025).
Comparison of AI Approaches.
Feature
Existing AI solutions
Proposed AI-driven framework
Target
Large corporations
Small municipalities, SMEs
Cost
High (custom AI models)
Low (cloud-based, modular)
Infrastructure
Requires strong digital base
Adaptable to low-digital areas
Scalability
Limited to large-scale rollouts
Highly scalable for local needs
Key Focus
Operational optimization
Bridging digital divide + optimization
Source: Prepared by the Authors (2025).
The framework is currently conceptual. We have outlined its components and logic, but it has not yet been fully implemented or tested in a real municipality. Full empirical validation will be a next step–likely through a pilot project in a willing small city or a simulated environment using historical data. In the meantime, we recognize certain potential limitations of the approach. Data sparsity is a concern: many small towns simply do not collect much data (e.g., they might not have IoT sensors or detailed records), which could limit the AI's effectiveness initially. We plan to mitigate this by designing the system to make the most of whatever data exists and by suggesting easy ways to gather more data (for example, starting a simple monthly log of certain metrics). Bias in AI models is another consideration–if the data is unrepresentative (perhaps only from one region or only from larger towns), the recommendations might not generalize. We address this by keeping humans in the loop to sense-check AI outputs and by possibly incorporating transfer learning from contexts like larger cities (with caution). Finally, scaling the prototype to many diverse communities could be challenging, given how heterogeneous local contexts are. A solution working in one town might need tweaking for another. We believe a cloud-based, modular system can allow such customization at low cost. A comparative summary of existing and proposed frameworks is presented in Table 2.
In implementing this framework, we can take advantage of existing platforms and tools to reduce development time and cost. There are several open-source libraries and cloud services suitable for our needs. For example, TensorFlow and PyTorch can be used to build and train the machine learning models (for clustering or RNN forecasting). Simpler tasks (like ARIMA forecasting) can be done with Python libraries like statsmodels. Cloud platforms like AWS SageMaker or Google Cloud AI Platform (now evolving into services like Google Vertex AI) provide managed environments to deploy these models, which is ideal for a small municipality that cannot maintain its own servers. Even without deep AI expertise, municipalities could use these platforms with pre-built templates. Additionally, partnerships with organizations like OpenAI (for advanced generative models) or national research institutions could provide access to cutting-edge AI (such as the use of large language models like ChatGPT to generate reports or answer questions for officials). The key is to use technology as an equalizer – enabling small entities to piggyback on the investments made by tech companies and research communities.
Before discussing broader comparisons, it's worth noting some real-world examples that inspired or informed elements of our framework. Globally, there are a few instances of AI being applied to circular or sustainability challenges at city/company level, which demonstrate the feasibility of our approach:
In Singapore, the national PUB has implemented a Circular Water management system where wastewater is recycled into high-quality potable water (the NEWater program). IoT sensors and AI analytics monitor water quality and usage patterns in real-time, optimizing the water treatment process and ensuring efficient water reuse. This high-tech solution addresses water scarcity and has reduced reliance on external water sources. It shows how data and AI can turn a waste product (sewage) into a resource (clean water). While Singapore is a wealthy city-state, the principles can be scaled down – for instance, a small city could use simpler sensors and AI models to detect leaks in its water supply, achieving water savings at low cost (Tortajada, 2024).
Caterpillar Inc. has a renowned remanufacturing program for heavy machinery components. Using data analytics and AI-driven predictive maintenance, Caterpillar identifies when engine parts or other components can be retrieved and remanufactured to as-new condition. This extends component lifespans and saves material costs on a large scale. The AI predicts part failures and optimal refurbishment timing, ensuring a steady supply of used parts for remanufacturing. For a municipality, the analogous idea is predictive maintenance of infrastructure – for example, using AI to predict when to refurbish equipment (boilers, vehicles) rather than buying new, thus saving money and resources in a circular fashion.
The city of Songdo, South Korea provides a case of AI-enabled waste management at an urban scale. Songdo has an automated waste collection system where garbage is sucked directly from kitchens through underground pipes to a central facility. There, AI-based sorting systems automatically segregate recyclables from general waste. This system reportedly led to a 40% reduction in waste sent to land (Shrivastava, 2024). While such infrastructure is beyond the reach of a small Polish town, components of it – like AI sorting – could be implemented in more modest ways (for example, regional recycling centers using AI vision to sort waste). The significant waste diversion achieved shows the potential impact of AI on circular economy goals.
In India, the government's Smart Cities Mission has encouraged several cities to adopt AI for urban services. For instance, the city of Bhopal implemented an AI-enabled waste collection and route optimization system that uses sensors on bins and GPS data to streamline garbage truck schedules. This led to more than a 30% improvement in collection efficiency and a notable reduction in fuel use for waste transport (as reported by the city) (Ictpost, 2025). This example illustrates that even in emerging economy contexts, with supportive policy and some investment, AI can deliver practical improvements in municipal services. It aligns with the needs of Polish municipalities to do more with limited resources.
These examples, along with others from Europe and around the world, show that the building blocks of our framework–data-driven clustering of problem areas, forecasting of needs, and optimization of operations–have proven effective in analogous scenarios. Our contribution is tailoring and combining these approaches specifically for under-resourced localities to create a holistic support system for circular economy planning.
Global Context and Comparative Analysis
To contextualize our findings, it is important to compare the identified barriers and proposed solutions with global trends in AI adoption for circular economy practices. This comparative analysis highlights both shared challenges and unique regional opportunities, offering a broader perspective on the scalability of the proposed framework. Our findings support the hypothesis that AI-driven approaches have significant potential for enhancing circular enterprises. However, they simultaneously highlight that the digital divide creates uneven adoption of these technologies and limits the potential impact of these technologies on the circular enterprises. For instance, the survey results indicate that over 70% of small cities identified financial constraints as their main challenge for the development of sustainable and circular practices. This result aligns with the bibliometric findings, that showed a lack of research focusing on smaller communities in relation to smart circular economy development. These findings highlight the need for targeted support and resource allocation for small communities to facilitate the adoption of circular economy practices and AI-driven solutions. Below is presented comparison table of the proposed AI-driven framework with existing AI-driven circular economy solutions:
Bibliometric analysis indicates that while countries such as India and Singapore have rapidly adopted AI-driven circular economy practices through strong public-private partnerships, Polish municipalities still face systemic budgetary and infrastructural barriers. In India, for instance, the “Smart Cities Mission” actively integrates AI for sustainable waste and resource management, a strategy not yet fully realized in Polish local governments.
The gap in AI adoption is also influenced by educational and digital infrastructure disparities. In contrast to Singapore's nationwide digital literacy initiatives, Polish small municipalities often lack targeted training programs for public administrators, limiting the scalability of AI-based solutions. Addressing these gaps is essential to foster equitable development in the context of smart circular enterprises.
In particular, our comparative analysis highlights a few critical points:
Financial and infrastructure gaps. The fact that 70% of Polish small cities cite funding as the main barrier aligns with experiences in many developing regions. It indicates that innovative financing models are needed. Internationally, we see green bonds, climate funds, or development bank loans being used to fund circular economy projects in cities. Poland's municipalities might explore similar avenues, potentially with support from EU funds earmarked for digital and green transformation. The proposed AI framework itself could help by identifying highest ROI (return on investment) projects to invest in first, making the case for funding more compelling.
Policy support. Countries that have made strides typically have strong policy frameworks. China's Circular Economy Promotion Law (2009) is an example of a national law driving local action. The EU's own Circular Economy Action Plan (2020) sets targets and provides support across member states. Poland can leverage EU policy support but also needs to translate it into on-the-ground action. The lack of focus on small communities in research suggests perhaps a similar lack in policy–policies often target big industries or cities. A shift in policy to explicitly include rural and small-town circular initiatives (with allocated resources) is necessary to ensure equitable development.
Technology transfer. Globally, there's a trend of technology transfer from advanced smart cities to smaller ones. For example, the AI waste sorting tech used in Songdo could be packaged and offered to smaller cities in other countries as a product or service. This is already happening with some cleantech companies. Polish municipalities could benefit from being part of international networks or pilot programs that bring proven solutions to them. Our framework, once pilot-tested, could similarly be shared beyond Poland, as many regions in Eastern Europe, Southeast Asia, Africa, etc., have comparable needs.
Scalability of AI solutions. A theme in global cases is scalability. Many existing AI solutions for circular economy are bespoke and implemented at large scale (city-wide or corporate-wide), which can be limited to those specific contexts. Our framework is designed to be highly scalable downwards–it can be deployed in a single small town–but also scalable across many towns by using cloud infrastructure. This approach of one framework serving many users is analogous to how some countries deploy centralized digital platforms for all municipalities. For instance, India's Ministry of Housing developed a common dashboard for tracking city performance under its programs. A similar approach could see our framework (or its future iteration) serving as a central AI hub for all interested Polish municipalities, with each seeing only their data, but benefiting from shared improvements and maintenance. This would drastically lower the per-town cost and maintenance burden.
The global perspective reinforces that while technology is a powerful enabler, it must be accompanied by institutional and policy support. Developed nations and forward-looking developing nations are investing heavily in both technology (AI, IoT) and human capital to ensure that sustainability goals can be met. Poland's situation is unique in that it has access to EU support and a strong educational base, yet internal disparities slow down uniform progress. By learning from international successes (like Ireland's community-level circular projects, or Singapore's digital upskilling programs) and failures, Poland can chart a path that leverages AI smartly.
In conclusion, our comparative analysis underscores that the hypothesis we address is not just a local issue but a global one: AI can indeed significantly enhance circular economy outcomes, but without deliberate action to include all communities, a digital (and sustainability) divide will persist. Our proposed framework, in concert with appropriate policies, aims to ensure that small municipalities are not left behind in the circular economy transition. The next section provides concrete policy and practical recommendations to realize this vision.
Policy and Practical Recommendations
Bridging the digital divide and enhancing AI adoption in circular economy enterprises will require coordinated policy interventions and on-the-ground strategies. Based on our findings and the global insights above, we outline the following key recommendations for policymakers, industry leaders, and local governments:
Subsidized access to cloud AI tools. National or regional governments should provide funded access to cloud-based AI platforms for small municipalities and SMEs. For example, a government program could cover the subscription costs for services like AWS, Microsoft Azure, or Google Cloud AI for qualifying towns/businesses, removing the financial barrier to using advanced AI tools (Brockman, 2019). This way, a small town can use high-powered analytics or machine learning models without capital expense. Such subsidies could be tied to specific goals (e.g., waste reduction targets) to ensure they are used effectively.
Targeted digital literacy and AI training programs. Stakeholders should invest in capacity-building initiatives for local officials and SME managers focused on digital skills and AI literacy. Creating specialized training modules (potentially in partnership with universities or online platforms) can equip local government employees with the knowledge to utilize AI tools and interpret their outputs. This could include short courses on data management, how to ask the right questions of an AI system, or how to integrate AI recommendations into planning. By improving digital literacy, we ensure technology solutions are actually adopted and maintained.
Pilot projects and funding for smart circular initiatives. To demonstrate feasibility and build momentum, allocate grant funding for pilot projects that apply AI to resource optimization in small towns or rural areas. These pilots could be competitions or targeted calls (for instance, funding 10 pilot municipalities to implement the AI framework for waste management and measure results). Successful pilots will create case studies that can be replicated. They also provide learning opportunities to refine tools like our framework. Importantly, pilots de-risk innovation for small entities – they can try new approaches with financial support and guidance, which if successful, can then attract further investment or scale-up funding.
Public-private partnerships. Encourage PPPs where tech companies or startups work with municipalities on circular economy challenges. For instance, a startup specializing in AI-based recycling could partner with a cluster of towns, providing solutions at a discounted rate in exchange for data or a testbed. The government can facilitate these matches and perhaps provide co-funding or incentives. This taps into the innovation and efficiency of the private sector while addressing public needs.
Infrastructure for data and connectivity. Although not the primary focus of our study, it became clear that some municipalities lack even basic digital infrastructure. Thus, national initiatives to improve broadband internet access and IT infrastructure in lagging areas are fundamental. Without internet connectivity or reliable computers, AI tools (especially cloud-based ones) cannot be effectively used. This ties into broader digital divide efforts beyond the scope of circular economy alone.
Regulatory support and simplification. Simplify and streamline regulations for implementing circular solutions and using digital tools. For example, ensure that there are clear guidelines and perhaps fast-track approvals for projects like smart waste facilities or data sharing agreements. If municipalities want to share data with a cloud service, clear legal frameworks should support that while protecting privacy. Reducing bureaucratic hurdles encourages local leaders to pursue innovative projects rather than being discouraged by red tape.
Community engagement and education. At a practical level, local governments should complement any AI or tech adoption with community engagement. For instance, if our AI framework identifies that a lack of recycling is an issue, the municipality should run awareness campaigns and involve citizens in co-creating solutions. Technology can pinpoint problems and suggest fixes, but human behavior change is often required to realize benefits (e.g., residents need to follow new waste sorting rules). Thus, coupling AI insights with strong communication and education efforts will yield the best outcomes.
These recommendations are interrelated and should be pursued in parallel. Together, they create an ecosystem in which a tool like our proposed AI framework can thrive and make a measurable impact. By lowering cost barriers, increasing knowledge, and providing supportive policy and funding, we can accelerate the integration of AI into circular economy initiatives for communities of all sizes. Further research can explore specific strategies for each of these recommendations – for example, the optimal structure of an AI training program for municipal staff, or the economic impact of cloud service subsidies – but there is enough evidence to begin implementation.
Conclusions and Future Work
This study provided both empirical evidence and a conceptual framework for advancing AI-driven circular economy practices, with a particular focus on small municipalities and bridging the digital divide in sustainable development. Through a nationwide survey of local government units, we found that while awareness of circular economy principles exists, implementation is hindered by financial, technological, and social barriers. Notably, about 70% of small municipalities indicated that funding constraints are the primary obstacle to adopting sustainable and circular practices, underlining the critical need for affordable and accessible solutions. The survey also revealed that the use of advanced digital tools like AI is currently very limited in these communities, which confirms our initial concern that a digital divide contributes to uneven sustainability outcomes.
The proposed AI-based framework addresses these findings by offering a low-cost decision support system aimed at improving resource efficiency and waste reduction at the local level. By leveraging readily available data and cloud-based algorithms (clustering, time-series forecasting, etc.), the framework can provide actionable insights to municipal decision-makers who lack in-house analytics capabilities. The framework's novelty lies in tailoring AI to the constraints of small municipalities: it is scalable, modular, and does not require significant upfront investment, which makes it feasible for widespread adoption among resource-limited entities. In doing so, the research supports our hypothesis that AI-driven approaches have significant potential to enhance the performance of circular enterprises, provided that issues of access and capacity (the digital divide) are consciously addressed. In fact, the study underscores that addressing the digital divide is not peripheral but central to achieving equitable sustainability transitions.
Our findings also highlight the importance of context-specific strategies. Challenges faced by Polish local governments–such as brain drain and lack of public awareness–indicate that technology alone is not a silver bullet. Hence, the integration of AI solutions must go hand in hand with community engagement and supportive policies (as outlined in our recommendations). One positive outcome of this research is identifying that many local officials are eager for solutions and willing to adopt new tools if given proper support. This human factor bodes well for the future implementation of the framework.
The contributions of this work are threefold: (1) It adds to the academic literature by clearly delineating the gap in research and practice regarding AI for circular economy in small communities (backed by our bibliometric evidence that “smart circular enterprise” research is scarce). (2) It provides a practical framework that stakeholders can build upon – essentially a blueprint for an AI tool that can be developed and tested in pilot projects. (3) It offers policy insights by connecting on-the-ground challenges with higher-level strategies needed to overcome them.
However, this study also has limitations that must be acknowledged. The framework proposed is conceptual and has not yet been validated with real-world performance data. Its effectiveness will ultimately need to be confirmed through pilot implementations. We have identified potential limitations such as data sparsity and model biases; future work should empirically test these and refine the system accordingly. Additionally, the survey data, while extensive, is self-reported and may contain biases (e.g., officials might underreport certain issues or overreport their progress). Future research could complement our survey with on-site audits or case studies for a more objective assessment.
Looking ahead, there are several avenues for future research and development:
Pilot and Proof-of-Concept Testing. The next immediate step is to implement the AI-driven framework in one or more municipalities as a proof of concept. This will involve software development (possibly in partnership with a tech firm or university department) and then training local users to integrate it into their workflow. Key metrics (the KPIs we outlined) would be tracked to evaluate impact. A pilot will provide valuable feedback on the framework's usability, accuracy, and impact, allowing us to iterate on the design. It will also surface any practical challenges in data collection or integration with existing systems.
Expanded Data Analysis. With more time and resources, one could enrich the data inputs to the framework. For example, incorporating satellite data for environmental monitoring or using mobile phone data to gauge population flows could enhance the analysis. Future studies might explore how unconventional data sources can improve local-level circular economy planning (e.g., using remote sensing to identify illegal dumpsites and then targeting them with interventions).
Inferential Statistical Analysis of Survey Data. As noted, our survey results invite deeper analysis. Future research should apply inferential statistics to test hypotheses such as “Municipality size is significantly correlated with the likelihood of having at least one AI initiative” or “Financial constraint mentions are independent of region when controlling for size.” Such analysis can help generalize findings beyond descriptive statistics. Moreover, a follow-up survey in a few years, after some interventions, could measure how the situation changes over time (a longitudinal study).
Interdisciplinary Integration. Our work mainly focused on technological and administrative aspects. There is room to integrate social sciences – for instance, studying how local culture or politics affects the adoption of AI in public services. Also, exploring economic models to quantify the benefits of circular economy improvements (like the monetary value of reducing landfill usage) can strengthen the case for these initiatives to stakeholders.
Scaling the Framework Beyond Poland. While our study centered on Poland, the framework and findings have broader relevance. Future research could test the approach in other countries, adjusting for local conditions, to verify its adaptability. This could also involve comparing results: does a small town in Poland benefit similarly from the framework as one in, say, India or Kenya? Comparative studies would be insightful.
In closing, this research reaffirms the potential of combining AI and circular economy efforts to create smarter, more sustainable communities. At the same time, it brings attention to the nuanced challenges of digital inequality. By implementing the recommendations and pursuing further research as outlined, policymakers and practitioners can move toward a future where even the smallest municipality can harness cutting-edge technology to build a sustainable, circular local economy. The journey from conceptual framework to real-world impact is just beginning, but the urgency of global sustainable development goals and the promising results thus far suggest that this is a journey worth undertaking.
Footnotes
ORCID iDs
Iurii Ganushchak
Ivana Ognjanović
Iryna Bashynska
Joanna Duda
Rafał Kusa
Funding
This contribution is the result of the research project co-financed by the European Union under the Horizon Europe Framework Programme (HORIZON-MSCA-2021-SE-01-1;Project number: 101086381;Project title: “Overcoming Digital Divide in Europe and Southeast Asia”),Polish Ministry of Science and Higher Education under Program “Projects International Co-financed” (“Projekty Międzynarodowe Współfinansowane”;agreement number: 5319/HE/2022/2023/2) and by program “Excellence initiative – research university” for the AGH University of Krakow (Action 21 nr. 8475).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.
References
1.
BarkemeyerR.HoltD.PreussL.TsangS. (2014). What happened to the ‘development’ in sustainable development? Business guidelines two decades after Brundtland. Sustainable Development, 22(1), 15–32. https://doi.org/10.1002/sd.521
BashynskaI.DyskinaA. (2018). The overview-analytical document of the international experience of building smart city. Business: Theory and Practice, 19169), 228–241. https://doi.org/10.3846/btp.2018.23
ChoubeyS.KarmakarG. P. (2021). Artificial intelligence techniques and their application in oil and gas industry. Artificial Intelligence Review, 54(04), 3665–3683. https://doi.org/10.1007/s10462-020-09935-1
9.
Czemiel-GrzybowskaW.Pavlakova-DocekalovaM.Ratajczak-MrozekM.ThompsonC. (2024). Sustainable development of smart cities through municipal waste incinerators: The examples of artificial intelligence in technological entrepreneurship.Engineering Management in Production and Services, 16(4), 127–138. [Source:] https://doi.org/10.2478/emj-2024-0037 (accessed 15 April 2025).
FilyppovaS.BashynskaI.KholodB.ProdanovaL.IvanchenkovaL.IvanchenkovV. (2019). Risk management through systematization. Risk Management Culture, International Journal of Recent Technology and Engineering, 8(3), 6047–6052. https://doi.org/10.35940/ijrte.C5601.098319
15.
GanushchakI. (2025). (Nie)zrozumieć korpoludka: Sekretny język korporacji IT. Poznań.
16.
HerlinaS.KusumahY.JuandiD. (2024). Global research on emerging digital technology: A bibliometric analysis.Journal of Advanced Research in Applied Sciences and Engineering Technology, 51(2), 281–294. https://doi.org/10.37934/araset.51.2.281294
HuangD.ZhanY.LonsdaleC. (2024). AI, the firm and the circular economy: Towards a maturity model. Academy of Management Proceedings, 2024(1). https://doi.org/10.5465/AMPROC.2024.17553
20.
ImranS.AlamK.BeaumontN. (2014). Reinterpreting the definition of sustainable development for a more ecocentric reorientation. Sustainable Development, 22(2), 134–144. https://doi.org/10.1002/sd.537
KirchherrJ.ReikeD.HekkertM. (2017). Conceptualizing the circular economy: An analysis of 114 definitions. Resources, Conservation and Recycling, 127(3), 221–232. https://doi.org/10.1016/j.resconrec.2017.09.005
23.
KristoffersenE. (2020). The smart circular economy: A digital-enabled circular strategies framework for manufacturing companies. Journal of Business Research, 120(4), 241–261. https://doi.org/10.1016/j.jbusres.2020.07.044
24.
KuangL.HeL. I. U.YiliR. E. N.KaiL. U. O.MingyuS. H. I.JianS. U.XinL. I. (2021). Application and development trend of artificial intelligence in petroleum exploration and development. Petroleum Exploration and Development, 48(1), 1–14. https://doi.org/10.1016/S1876-3804(21)60001-0
25.
LewickaD.Glińska-NeweśA.BatkoR.BashynskaI.Ur RehmanH. (2025). Organizational support for the circular economy through shaping employee green behaviour. Edward Elgar Publishing Ltd., 250.
26.
LiX.WangZ.ChenC.ZhengP. (2021). A data-driven reversible framework for achieving sustainable smart product-service systems. Journal of Cleaner Production, 279(2021), 30. https://doi.org/10.1016/j.jclepro.2020.123618
27.
McCarthyJ.MinskyM. L.RochesterN., et al.A proposal for the Dartmouth summer research project on artificial intelligence. [Source:] http://jmc.stanford.edu/articles/dartmouth.html (accessed 15 September 2024).
28.
MeadowsD. H.MeadowsD. L.RandersJ. (1992). Beyond the Limits: Global Collapse or a Sustainable Future. Chelsea Green Publishing.
29.
MeadowsD. H.MeadowsD. L.RandersJ.BehrensW. W. (1972). The limits to growth: A report for the club of Rome’s project on the predicament of mankind. Potomac Associates – Universe Books.
30.
MeadowsD. H.RandersJ.MeadowsD. (2004). Limits to Growth: The 30-Year Update. White River Junction.
31.
MertonR. K. (1967). On theoretical sociology: Five essays, old and new. Free Press.
NižetićS. (2019). Smart technologies for promotion of energy efficiency, utilization of sustainable resources and waste management. Journal of Cleaner Production, 231(2019), 565–591. https://doi.org/10.1016/j.jclepro.2019.04.397
34.
OchiengE.OmindeD.ZuofaT. (2024). Potential application of generative artificial intelligence and machine learning algorithm in oil and gas sector: Benefits and future prospects. Technology in Society, 79(2). https://doi.org/10.1016/j.techsoc.2024.102710
35.
PathanM. S.RichardsonE.GalvanE.MooneyP. (2023). The Role of Artificial Intelligence within Circular Economy Activities – A View from Ireland. Sustainability, 15(12), [Source:] https://doi.org/10.3390/su15129451 (accessed 22 October 2024).
United Nations, Agenda 21: Programme of Action for Sustainable Development. United Nations Conference on Environment and Development. Rio de Janeiro 1992.
44.
World Commission on Environment and Development (WCED). (1987). Our Common Future.Oxford University Press.
45.
Zgrzywa-ZiemakA. (2019). Model zrównoważenia przedsiębiorstwa. Wroclaw.
