Smart Innovation for a Circular Economy: A Systematic Review of Emerging Trends and the Future of AI in the Sustainable Economy

Abstract

The present study explores the potential of Artificial Intelligence (AI) in driving the sustainable economy forward through the optimization of resources and the mitigation of environmental impacts. However, the study also identifies significant challenges in the conceptual and methodological integration necessary to establish a comprehensive theoretical framework. The heterogeneity of these approaches hinders the discernment of discernible patterns within the scientific literature, thereby impeding the development of efficacious strategies to optimize their impact on sustainability. The present study employs the PRISMA 2020 methodology to analyze trends in scientific production on AI and the sustainable economy, thereby ensuring a rigorous process of identifying and evaluating studies. The mapping of the academic community, the geographical distribution of knowledge, and the evolution of approaches allow us to understand the research dynamics and identify gaps in the literature. This study makes a significant contribution to the field by offering a comprehensive perspective that supports the integration of AI into sustainable models and its application in sectors with high environmental and economic impacts.

1. Introduction

The advent of Artificial Intelligence has precipitated a paradigm shift within the economic and social sectors, thereby engendering a state of profound transformation. The capacity to process data, enhance operational efficiency, and generate predictions renders it applicable across diverse sectors, including those associated with sustainability [1]. In this domain, AI plays a pivotal role in fostering a sustainable economy by optimizing resource utilization, enhancing decision-making processes, and mitigating adverse environmental impacts [2].

It is evident that smart AI-based innovations have the potential to address environmental and economic challenges through the implementation of advanced solutions. The utilization of tools such as machine learning and real-time data analytics has been demonstrated to optimize supply chains, reduce waste, and enhance energy management [3]. Moreover, in the domains of precision agriculture and sustainable mobility, AI plays a pivotal role in curtailing the excessive utilization of inputs and promoting more environmentally responsible practices [4].

A comprehensive analysis of prevailing research trends in the fields of artificial intelligence and the sustainable economy is imperative. The identification of patterns in scientific production facilitates the comprehension of the evolution of knowledge, facilitates recognition of key academic contributions, and provides a framework for the guidance of future research [5]. A structured approach to the current state facilitates strategies to strengthen the relationship between AI, innovation, and sustainability, promoting balanced and environmentally responsible economic development.

The study of the uses of Artificial Intelligence in the sustainable economy is fragmented due to the diversity of approaches, methodologies, and applications in the scientific literature. This dispersion hinders the establishment of unified knowledge, which is essential for the development of a solid foundation for future research. Despite the growing interest in the relationship between AI and sustainable innovation, the absence of a coherent structure in academic production hinders the identification of connections between studies and the evaluation of their impact [6,7].

The absence of a structured organization in the extant literature hinders the identification of key patterns and trends in this field. In the absence of a comprehensive understanding of research advancements and lacunae, it becomes challenging to identify the primary references, the most pertinent contributions, and the emergent trajectories of knowledge. Consequently, the development of strategies that optimize the potential of AI for sustainability is limited [8,9].

This situation underscores the necessity for a thorough examination of scientific output, which facilitates the discernment of the dynamics of knowledge in this domain and the enhancement of the framework governing the relationship between AI and the sustainable economy.

In this sense, the objective of this research is to identify research trends in the use of Artificial Intelligence for the sustainable economy by means of analysis of scientific production, the main academic references, geographical distribution, and emerging keywords. In order to achieve this objective, a series of questions must guide the analysis of scientific production in this field.

• How many scientific publications on Artificial Intelligence and the Sustainable Economy have been generated annually?
• Who are the main authors and journals publishing in this field?
• What are the most cited studies and their impact on the discipline?
• How is scientific production on this topic distributed globally?
• What are the emerging, growing, and declining keywords in this line of research?

The research questions were formulated based on the identified problem and the stated objective of the study. In light of the fragmented and dispersed nature of extant literature on artificial intelligence and the sustainable economy, the research questions endeavor to provide a structured understanding of the field by focusing on key dimensions such as the volume and evolution of scientific output, the most influential contributors, geographical trends, citation impact, and the dynamics of keyword usage. The aforementioned inquiries were meticulously crafted to align with the overarching research objective, which was to identify trends in the utilization of artificial intelligence for sustainable economic development. This objective was pursued with the dual aims of facilitating the organization of knowledge and recognizing existing gaps in research. By doing so, the study seeks to provide guidance for the future direction of research in this field.

In conclusion, the present study makes a significant contribution to the existing body of knowledge concerning the utilization of artificial intelligence within the context of the sustainable economy. This contribution is evidenced by the comprehensive analysis of the extant academic literature in this field. The updating of knowledge is essential for the identification of gaps in literature and the consolidation of an information base that facilitates future research. The identification of patterns, influential authors, and emerging trends provides a comprehensive overview that can guide research efforts. In this manner, the study not only organizes existing knowledge but also establishes the foundation for a research agenda that responds to current and potential challenges at the intersection of AI, innovation, and sustainability.

2. Materials and Methods

The research adopts the PRISMA 2020 methodology, a fundamental guideline for systematic reviews that guarantees transparency and reproducibility [10]. This approach is predicated on the structuring of the process into phases of study, identification, selection, evaluation, and synthesis, thus ensuring a rigorous collection of evidence. The application of this study facilitates analysis of the utilization of artificial intelligence in the sustainable economy within the context of smart innovations, thereby consolidating extant literature and identifying knowledge gaps. The analysis is based on objective criteria and a robust methodological framework.

2.1. Eligibility Criteria

The selection of studies was based on inclusion and exclusion criteria that guaranteed the relevance and quality of the information. The research encompassed studies on artificial intelligence applied to sustainable economic models, with a particular emphasis on machine learning and AI-based solutions for sustainability and the circular economy. A variety of studies were considered for this research, including empirical evidence, theoretical models, and relevant systematic reviews.

The exclusion process was structured into three phases. The initial filtering of documents revealed indexing errors, i.e., incorrect categorization in the consulted databases, which led to a reduction in the presence of studies outside the thematic scope. The second evaluation assessed the availability of full-text access; however, no exclusion criteria were applied at this stage because only studies with full access were analyzed, ensuring verifiable and complete information. The third article was eliminated on the basis that it did not demonstrate a direct relationship to the subject under discussion. Research on artificial intelligence that was not related to the sustainable economy, as well as studies on sustainability or the circular economy that did not incorporate AI, were excluded from consideration.

2.2. Sources of Information

The research was grounded in documents indexed in Scopus and Web of Science, two scientific databases that provide access to relevant and validated information. Scopus is a multidisciplinary database that contains peer-reviewed publications in the fields of technology, economics, and sustainability. Its comprehensive coverage facilitates the identification of studies on artificial intelligence applied to sustainable economic models, ensuring a rigorous review process in the selection of scientific literature [11].

Web of Science is a bibliographic database that indexes high-impact scientific journals and provides access to essential research in the fields of innovation and sustainability. The structure of the database enables advanced searches and the selection of documents based on their relevance and methodological rigor. This contributes to strengthening the quality of the analysis and the validity of the results [11].

2.3. Search Strategy

In order to identify the studies included in the research, a specific search equation was defined for each database, in accordance with the inclusion criteria. In the Scopus database, the following equation was used: **TITLE (“Artificial Intelligence” OR “AI” OR “machine learning” OR “AI-based solutions” OR “deep learning”) AND TITLE (“Circular Economy” OR “sustainable economy”)**. In Web of Science, the syntax was adapted to **TS = (“Artificial Intelligence” OR “AI” OR “machine learning” OR “AI-based solutions” OR “deep learning”) AND TS = (“Circular Economy” OR “sustainable economy”)**.

The data corresponding to the year 2025 were collected up to 15 May 2025, based on the availability of indexed publications in Scopus and Web of Science at the time of data retrieval. In order to ensure a valid comparison with full-year data from previous years, the number of publications in 2025 was normalized using a linear projection, under the assumption of a constant publication rate throughout the year. This adjustment enables an equitable comparison of trends in Figure 1, thereby avoiding the underrepresentation of 2025 due to partial-year data. All information related to the included studies and the selection process is available in the Supplementary Material.

2.4. Selection Process

The selection of studies followed a structured process based on the PRISMA flowchart (Figure 1). Initially, documents were identified by employing the search equation utilized in Scopus and Web of Science. Duplicate records and documents containing indexing errors were then removed in order to avoid inconsistencies in the data collection process.

Titles and abstracts were then reviewed to assess the relevance of each article according to the inclusion criteria. A comprehensive analysis of the preselected documents was subsequently undertaken to ascertain their congruence with the study’s objectives.

This procedure guaranteed a rigorous and transparent selection of information, ensuring the inclusion of only relevant studies in the final analysis.

In order to address the concerns regarding the relatively low number of articles retrieved (fewer than 100), it is imperative to clarify that the selection process prioritized thematic specificity and methodological rigor over volume. The search strategy, implemented in the title fields of both Scopus and Web of Science, was deliberately centered on core terms—“Artificial Intelligence,” “machine learning,” “AI-based solutions,” “circular economy,” and “sustainable economy”—to ensure that the resulting documents directly aligned with the study’s objective. This methodological decision served to minimize the inclusion of articles with tangential relevance, thereby enhancing the conceptual coherence of the review.

The paucity of extant literature on this subject is indicative of the emergent and interdisciplinary nature of this research area, which is situated at the intersection of advanced technologies and sustainable development. As indicated by recent bibliometric analyses [12,13,14], this subject has been comparatively understudied in relation to more established domains, a factor that accounts for the paucity of consolidated scientific output. The novelty of the field is a contributing factor to its niche status, and it reinforces the importance of conducting structured and focused reviews to build foundational knowledge.

2.5. Data Processing

Microsoft Excel was utilized as the primary instrument for data management and analysis. A structured database was created with the selected articles, organized by year of publication, country of origin, methodology used, and main findings. The application of Excel filters and segmentations was utilized to identify patterns and trends in the reviewed literature. Furthermore, the utilization of pivot tables and graphs facilitated the visualization of the results in a manner that was both clear and comprehensible. This methodology enabled the efficient organization of information and facilitated comparative analysis between studies, ensuring systematic and rigorous processing of the data obtained.

2.6. Risk of Bias

In order to reduce the risk of bias, strategies were implemented to guarantee the quality and relevance of the analyzed studies. The selection of literature was conducted using recognized databases, thereby reducing the inclusion of literature with low rigor. However, the utilization of these sources has the potential to exclude non-indexed publications, thereby affecting the diversity of approaches.

The inclusion and exclusion criteria were established to ensure thematic consistency; however, the selection of key terms may have influenced the information retrieval process and restricted certain approaches. Furthermore, the phenomenon of reporting biases, which are derived from the tendency to publish positive or significant results, was given due consideration. The selection process incorporated systematic validation to mitigate the aforementioned effects and enhance the objectivity of the analysis, thereby ensuring the reliability of the findings.

Additionally, this study deliberately excluded grey literature, including conference proceedings, technical reports, and non-peer-reviewed sources. Although these documents may offer valuable insights, especially in rapidly evolving fields such as artificial intelligence, their inclusion was not considered appropriate due to the lack of rigorous peer review processes that ensure methodological quality and reliability. The decision to focus exclusively on peer-reviewed journal articles was made to maintain a high standard of scientific rigor and to minimize the potential inclusion of unvalidated or preliminary findings. Future research could explore complementary analyses that integrate grey literature to capture emerging developments not yet consolidated in the formal academic publication system.

3. Results

A significant increase in scientific output pertaining to the utilization of artificial intelligence within the context of the sustainable economy has been observed from 2019 to 2025. As demonstrated in Figure 1, the number of publications increased from two in 2019 to six in 2020 and seven in 2021. This figure rose further to 13 in 2022, 19 in 2023, and 38 in 2024. By 2025, a total of six publications are projected. This trend exhibits a linear growth coefficient of determination (R²) of 0.8588, indicative of a robust model fit and a substantial relationship between time and scientific output. Figure 2 illustrates this growth, reflecting an increasing interest in artificial intelligence applied to the sustainable economy, driven by the necessity for intelligent innovations to address environmental and productive challenges.

The analysis of the leading authors in the literature on artificial intelligence and the sustainable economy reveals three distinct groups, as illustrated in Figure 3. The initial group, denoted by the color orange, comprises researchers who have exerted the most significant influence in terms of citations and publications. However, no authors were identified in this category in the present study.

The second group, represented in blue, comprises authors with a limited number of publications but a high number of citations in comparison to the average. This category comprises Bag S, Gupta S, Dwivedi YK, Pretorius JHC, Lopes SI, Fernández-Caramés TM, and Fraga-Lamas P. The third group, represented in green, includes researchers who demonstrate high academic productivity yet receive a low number of citations. The group under discussion includes Seyyedi SR, Gheibi M, Kowsari E, Chinnapan A, Pathan MS, Ali ZA, Zain M, Wang Y, Mooney P, Soo A, Ramakrishna S, and Shon HK.

Moreover, an analysis of the leading journals in the field of artificial intelligence and sustainable economics reveals the presence of three distinct groups, as illustrated in Figure 4. The initial group, denoted by the color orange, comprises the journals with the highest impact in terms of citations and the number of publications. Notably, Sensors, Journal of Cleaner Production, Process Safety and Environmental Protection, and Sustainability are particularly prominent. The journals under discussion have consolidated their relevance in the field by hosting a significant volume of studies and receiving high recognition in the scientific community.

The second group, represented in blue, comprises journals with a low number of publications but a high number of citations in comparison to the mean. This category includes Technological Forecasting and Social Change, which, despite their lower publication frequency, have a notable influence on the discipline, reflecting the relevance of the studies they host. The third group, represented in green, comprises journals with high academic productivity but a lower number of citations. The group under discussion includes the following publications: the Journal of Environmental Management, Environment, Development and Sustainability, Desalination, Heliyon, Recycling, and Circular Economy and Sustainability. These journals have made a significant contribution to the scientific output in this field, although the extent of their impact is less evident in terms of citations received. This suggests potential avenues for enhancing the influence of these journals within the academic community.

A thorough examination of the ten most frequently cited articles in the extant literature on artificial intelligence and the sustainable economy has been undertaken in order to identify the studies with the most significant impact in the field, as illustrated in

Table 1. Main contributions in the literature.

Title	Authors	Citas
Role of institutional pressures and resources in the adoption of big data analytics powered artificial intelligence, sustainable manufacturing practices and circular economy capabilities	[15]	495
Green iot and edge AI as key technological enablers for a sustainable digital transition towards a smart circular economy: An industry 5.0 use case	[16]	219
Deep learning model for Demolition Waste Prediction in a circular economy	[17]	120
Artificial Intelligence in the Sorting of Municipal Waste as an Enabler of the Circular Economy	[18]	87
Environmental cost control system of manufacturing enterprises using artificial intelligence based on value chain of circular economy	[19]	86
From Trash to Cash: How Blockchain and Multi-Sensor-Driven Artificial Intelligence Can Transform Circular Economy of Plastic Waste?	[20]	83
Artificial intelligence-based solution for sorting COVID related medical waste streams and supporting data-driven decisions for smart circular economy practice	[21]	72
The circular economy meets artificial intelligence (AI): understanding the opportunities of AI for reverse logistics	[22]	65
Machine learning approach for a circular economy with waste recycling in smart cities	[23]	61
Improving energy efficiency of carbon fiber manufacturing through waste heat recovery: A circular economy approach with machine learning	[24]	60

. The most referenced work is that of Bag et al. [15], with 495 citations, which examines the institutional pressures and resources required for the adoption of artificial intelligence in sustainable manufacturing and the circular economy. This is followed by the study by Fraga-Lamas et al. [16], which has received 219 citations, and which analyses the role of the Green Internet of Things and artificial intelligence in the digital transition towards a smart circular economy.

Akanbi et al. [17] developed a demolition waste prediction model in a circular economic context, with the model receiving 120 citations [14]. In their 2021 study, Wilts et al. [18] investigated the application of artificial intelligence in municipal waste sorting, with the study receiving 87 citations [18]. Chen et al. [19], with 86 citations, present an AI-based environmental cost control system, while Chidepatil et al. [20], with 83 citations, explore the use of blockchain and sensors to optimize plastic waste management.

In their 2021 study, Kumar et al. [21] propose an artificial intelligence solution for the sorting of medical waste related to the virus known as SARS-CoV-2, which causes the disease known as Coronavirus Disease 2019 (COVID-19). The study has been cited 72 times. In their 2022 study, Wilson, Paschen, and Pitt [22] conducted an extensive review of the extant literature on the impact of artificial intelligence on reverse logistics, citing 65 relevant sources in the process. Chen [23], with 61 citations, investigates the recycling of waste in smart cities, and Khayyam et al. [24], with 60 citations, address the issue of energy efficiency in the manufacturing of carbon fiber from a perspective that is consistent with the principles of the circular economy.

When analyzing the ten most cited articles presented in Table 1, it is essential not only to acknowledge their impact in quantitative terms but also to understand the qualitative reasons that explain their influence in the literature. A deeper examination of these studies reveals that their relevance is largely due to the methodological rigor they applied and the originality of the approaches they proposed. For example, the study by Bag et al. [15] integrates perspectives from institutional theory and resource-based theory, providing a solid framework for understanding how external pressures and resource availability influence the adoption of artificial intelligence in sustainable manufacturing. This rigorous methodological approach has facilitated the replication and expansion of their findings in subsequent research, which explains their high citation count.

Similarly, the work of Fraga-Lamas et al. [16] stands out for introducing a novel approach that combines Green IoT and Edge AI in the digital transition towards a smart circular economy. Their contribution goes beyond technological description by proposing an integrated architecture that optimizes industrial processes while considering energy and environmental challenges. Such methodological and conceptual innovations are what position these studies as key references in the field. The same applies to other articles like that of Akanbi et al. [17], which is notable for developing deep learning predictive models for managing demolition waste—a research line that had been previously underexplored.

Furthermore, studies such as Chidepatil et al. [20], which explore the combination of blockchain and smart sensors for plastic waste management, provide practical solutions with high real-world application potential. These articles not only introduce new technologies but also open discussions on how digitalization can transform circular business models. Collectively, the most cited articles contribute theoretical frameworks, robust methodologies, and disruptive applications that have enriched the understanding of the role of artificial intelligence in the sustainable economy. Therefore, their influence should not be interpreted solely by citation volume but also by the quality and relevance of their contributions to advancing knowledge in this field.

An analysis of the global distribution of research on artificial intelligence in the sustainable economy reveals a concentration in specific countries, as illustrated in Figure 5. Italy has the highest scientific output, with eight publications, followed by India, Spain, and the United Kingdom, which have each published seven studies. China has contributed five studies, while Canada, the United Arab Emirates, Finland, and Portugal have each published four. The Republic of Korea has contributed a total of three studies to the ongoing discourse on the aforementioned topic.

Europe is the leading continental contributor, with Italy, Spain, the United Kingdom, Finland, and Portugal being the primary contributors. Significant contributions are also made by Asia, with India, China, the United Arab Emirates, and South Korea reflecting a growing interest in the region for smart innovations applied to economic sustainability. North America is represented by Canada, while other regions demonstrate a comparatively lower level of research activity on this topic. This distribution indicates that the countries with the highest output have advanced technological and institutional capabilities for the development of studies in artificial intelligence and sustainability.

Moreover, an analysis of keywords in the extant literature on artificial intelligence and sustainable economy reveals trends based on their average year of use and frequency, as illustrated in Figure 6. The initial quadrant encompasses the terms that have been the subject of the most research, that are utilized most frequently, and that are undergoing the most rapid growth. The second quadrant encompasses emerging concepts, which, although less prevalent, have a recent average year of appearance. The fourth quadrant comprises concepts that are the focus of extensive research, yet have seen a decline in the frequency of their appearance in recent studies.

The initial quadrant encompasses the keyword “Sustainable Development,” thereby signifying its pertinence within contemporary research. The second quadrant, corresponding to emerging terms, identifies AI-Based Circular Economy, Algorithms, Anaerobic Digestion and Medium Enterprise, Smart Cities, and Sustainable Development Goals, thereby highlighting new lines of study in the relationship between artificial intelligence and sustainability. In the fourth quadrant, representing decreasing concepts, no keywords were identified in this analysis.

4. Discussion

The discussion has been organized into key sections in order to analyze the uses of artificial intelligence in the sustainable economy within the framework of smart innovations. Firstly, the results are presented and critically examined, structured within a conceptual framework that synthesizes the main findings of the literature. A subsequent comparison of these results with those of previous studies is made in order to identify both similarities and differences in the field. The subsequent stage of the research process involves the analysis of existing gaps in knowledge, highlighting aspects that remain to be explored or which have been addressed only in a limited capacity. It is proposed that a research agenda be formulated to guide future studies, in light of the identified trends and the existing gaps. Finally, the theoretical, policy, and practical implications of the findings are discussed, along with the methodological and scope limitations of the study.

4.1. Analysis of Results

A comprehensive review of the extant literature on the intersection of artificial intelligence and the sustainable economy reveals a discernible evolution in the theoretical development of the field. In the nascent stages of research in this field, the focus was on the exploration of approaches to the integration of artificial intelligence (AI) into sustainable models. For instance, the study by Sanakaran [25] provides a compelling link between AI and the circular economy, as well as the energy transition. However, academic productivity was constrained. The observed increase in publications between 2022 and 2024 signifies a theoretical maturation, marked by a transition from conceptual proposals to concrete applications. As demonstrated in the study by Singagerda et al. [26], the utilization of artificial intelligence (AI) in the domain of waste management is approached through the lens of an empirical and normative methodology.

A comprehensive review of the extant literature on AI and the sustainable economy reveals the presence of three distinct groups of researchers, each exhibiting unique patterns of citation and productivity. Bag et al. [15] conducted a study on the adoption of AI in sustainable manufacturing, emphasizing the impact of institutional pressures and resources. The present study’s findings lend support to the theoretical framework for the utilization of big data in the context of circular economies. In their 2023 study, So et al. conducted an analysis of machine learning in the context of urban nutrient recovery, with a particular focus on its impact on waste management and agricultural efficiency. The findings of both studies demonstrate the potential of artificial intelligence (AI) to enhance sustainability and the circular economy.

The analysis of journals on AI and sustainable economics employs a three-category classification system based on impact and productivity. A substantial body of research has demonstrated the significant impact and citations of both sensors and the Journal of Cleaner Production, as evidenced by studies such as Fraga-Lamas et al. [16] on Green IoT and Edge AI in industrial sustainability. The publication Technological Forecasting and Social Change has a relatively limited number of publications, yet it has received a high number of citations. For example, Bag et al. [15] investigated the adoption of big data in sustainable manufacturing.

The study by Bag et al. [15] has exerted a considerable influence on the field, and this can be attributed to its innovative integration of institutional theory and resource-based perspectives in the context of artificial intelligence and sustainable manufacturing. This combination provided a solid theoretical framework that has been widely adopted and referenced by subsequent studies seeking to explain how external pressures, resource availability, and organizational flexibility shape the adoption of artificial intelligence (AI) in sustainability-driven production systems. Furthermore, Bag et al. [15] addressed a critical gap by establishing a link between big data analytics and practical applications in circular economy practices. The researchers offered validated models that have served as a foundation for both empirical research and policy discussions. The elevated citation rate is indicative of two factors. Firstly, it is a testament to the methodological rigor of the work. Secondly, it demonstrates the work’s versatility for application across multiple sectors.

A similar argument can be made for the study by Fraga-Lamas et al. [16], which gained substantial academic impact by introducing an integrated vision of Green IoT and Edge AI as enablers of sustainable digital transformation. This study diverges from previous works in its analysis of these technologies. Previous studies analyzed these technologies in isolation; this study presents a cohesive architecture that connects industrial processes with sustainability objectives through intelligent, decentralized systems. This approach was developed in response to the mounting imperative for solutions that address the simultaneous imperatives of technological advancement, energy efficiency, and environmental concerns. The article’s influence is evident in its frequent citation across studies focused on Industry 5.0, smart cities, and sustainable supply chains. In these studies, the article is often referenced as a key conceptual and practical model to guide further technological integration. The recycling of materials is indicative of high productivity and low citations, including the work by Pregowska et al. [27] on AI in battery management.

A review of the extant literature on artificial intelligence and sustainable economics reveals that the most influential studies are those by Bag et al. [15] and Fraga-Lamas et al. [16] in terms of both impact and citations. Bag et al. [15] examined how institutional pressures and resources influence the adoption of AI in sustainable manufacturing, validating a theoretical framework based on institutional theory and the resource view. The results of the study suggest that organizational flexibility and industrial dynamism are critical factors in enhancing efficiency and circularity in production.

Fraga-Lamas et al. [16] analyze Green IoT and Edge AI as technologies that facilitate the sustainable digital transition. The results of the study suggest that these tools enhance the efficiency of industrial processes. However, they also present substantial energy and environmental challenges due to resource consumption and the impact of cloud computing. The two studies provide valuable insights into the potential and limitations of AI in the context of sustainability, thereby establishing a foundation for future research endeavors focused on energy efficiency and sustainable digitalization within the framework of circular economies.

The prevailing focus of research on artificial intelligence (AI) and the sustainable economy has been on countries with high levels of technological development. Italy is the leading nation in this regard. The studies by Roberts et al. [28] and Sabale et al. [29] analyze the relationship between AI and the circular economy from different perspectives. In their study, Roberts et al. [28] examined the role of artificial intelligence (AI) in the transition to a circular economy, highlighting the ethical risks involved and the need for regulations to mitigate negative impacts. The analysis conducted herein underscores the significance of achieving a balance between innovation and responsibility in sustainable models.

Sabale et al. [29] conducted a study on the utilization of artificial intelligence (AI) in micro, small, and medium-sized enterprises (MSMEs) within the context of Industry 4.0. The results of the study indicate that the implementation of artificial intelligence (AI) models has the potential to optimize efficiency, minimize expenses, and promote competitiveness within a circular economy framework. The two studies under consideration both emphasize the significance of a strategic framework that integrates technological advancements with ethical and sustainable principles, thereby positioning Italy as a leader in the fields of artificial intelligence, the circular economy, and digital transformation.

The analysis presented in Figure 5 determined the country attribution of each article based on the main institutional affiliation of the first author, as reported by the respective scientific databases (Scopus and Web of Science). In instances of co-authorship involving researchers from multiple countries, the primary affiliation listed for the publication was given precedence. This approach was adopted to avoid duplications in country counts and to maintain consistency in the geographical distribution analysis. Consequently, the total number of articles per country is indicative of the affiliation most prominently associated with the publication, rather than a cumulative count across all co-authors.

However, a review of global scientific output reveals that research on artificial intelligence applied to the sustainable economy is heavily concentrated in certain regions, while countries in Latin America, Africa, and parts of Southeast Asia show significantly lower participation. This discrepancy hinders the evolution of context-specific methodologies and curtails the global relevance of sustainability frameworks grounded in artificial intelligence. In the context of underdeveloped regions, impediments such as constrained infrastructure, inadequate financial resources allocated to digital innovation, and deficient regulatory frameworks impede advancements. For instance, even in technologically advanced countries like China, challenges in implementing smart waste management systems under a circular economy approach persist [20], while in Canada, limitations related to integrating circular principles into smart urban systems have been identified [30] illustrating that common barriers exist across both high- and low-output regions.

To address these gaps, it is imperative to foster North–South scientific collaborations. Such collaborations are essential for building local capacities, enabling knowledge transfer, and promoting co-created research adapted to diverse socio-economic realities. Recent studies from the United States on integrating circular economy principles into smart city frameworks [31] and from Spain on the development of sustainable smart mobility models [32] offer valuable insights that could serve as foundations for collaborative efforts with countries producing less scientific output. These international partnerships have the potential to facilitate the development of inclusive solutions that align technological advancement with ethical, social, and environmental principles, thereby contributing to a more equitable and globally relevant transition toward a digital and sustainable economy.

A study of keywords in the literature on artificial intelligence (AI) and the sustainable economy reveals the evolution of relevant terms. The notion of sustainable development is currently experiencing a period of heightened prominence, thereby indicating its growing significance in the context of the transition to circular economic models. Ronaghi [33] examines the manner in which artificial intelligence (AI) in manufacturing enhances resource management and optimizes sustainable practices. Among the emerging terms, “AI-Based Circular Economy” and “Smart Cities” indicate novel research directions in the domains of digitalization and urban sustainability. In their 2024 publication, Seyyedi et al. [34] underscore the potential of the Internet of Things (IoT) to enhance energy efficiency and facilitate strategic decision-making in the domain of waste management. The employment of artificial intelligence has been demonstrated to facilitate more efficient and sustainable circular economies.

As demonstrated in Figure 7, the consolidated framework synthesizes the primary research outcomes. The present scheme organizes the findings on the application of artificial intelligence in the sustainable economy, thereby highlighting its benefits, uses, and challenges. Furthermore, it facilitates the identification of connections between emerging concepts and trends. This provides a clear structure for analysis and decision-making in future studies.

4.2. Comparison with Other Studies

The analysis of the utilization of artificial intelligence in the sustainable economy within the context of smart innovations reflects trends identified in previous studies and reveals relevant differences.

The categorization of authors into three groups based on impact and productivity is consistent with the findings of previous bibliometric studies, such as that of Jrad [35], which also categorized influential authors in the fields of AI and economics. However, in contrast to the findings of Jrad, this study did not identify researchers in the category of highest impact in citations and publications, suggesting a less consolidated state in this specific area.

With regard to academic journals, the findings are consistent with those of Bracarense et al. [30], who noted that publications such as the Journal of Cleaner Production and Sustainability play a central role in the dissemination of knowledge on AI and sustainability. However, the present study expands the classification by including journals with lower impact but high academic productivity, suggesting greater diversity in research sources.

From a geographical perspective, the distribution of academic production by country aligns with the results of Zaidi et al. [13], who identified Europe and Asia as leaders in research on AI and sustainable economies. However, Bracarense et al. [12] have highlighted the paucity of research on the economic impact of AI on sustainability, a gap that has also been identified in this analysis.

A keyword analysis reveals similarities with Sulich et al. [14], who highlighted the relevance of AI in sustainable business decision-making. Whilst the study identified a decline in concepts, the present analysis did not identify any terms in that category. This suggests a focus on emerging trends such as AI-Based Circular Economy and Smart Cities.

As illustrated in

Table 2. Research gaps.

Category	Gap Identified	Future Research Question
Regulation	Lack of clear regulatory frameworks	How should AI regulations be designed in sustainable economies?
Ethics	Algorithmic biases in economic decisions	How to mitigate bias in AI applied to sustainability?
Business Models	Under development of circular AI models	Which AI-based business models support the circular economy?
Economic Impact	Limited evaluation of return on investment	How to measure the economic impact of AI on sustainability?
Accessibility	Barriers to adoption in SMEs	What strategies facilitate AI adoption in sustainable small businesses?
Infrastructure	Deficit in quality data for AI	How to improve data availability and quality for sustainable AI?
Energy Consumption	High energy cost of AI	How to optimize AI energy consumption in sustainability?
Interdisciplinarity	Limited integration of environmental sciences and AI	How to foster interdisciplinary approaches to AI for sustainability?
Global Inequality	Innovation gap between countries	How to reduce inequality in access to and development of sustainable AI?
Impact Measurement	Lack of standardized metrics	What indicators allow for an accurate assessment of AI in sustainability?

, there is a paucity of research in the use of artificial intelligence for the sustainable economy in the context of smart innovations. The table also presents questions that have been designed to address these research gaps in the literature. The construction of the program is predicated on the identification of areas that have not been extensively explored, with a view to the enhancement of scientific knowledge and the stimulation of new research directions.

4.3. Research Gaps

The identification of the research gaps presented in Table 2 is directly connected to the patterns and limitations observed in our own analysis of the scientific literature. For instance, the absence of regulatory frameworks and standardized impact measurement tools became evident during the review of the most cited articles, which often focused on technical applications without addressing the regulatory or evaluation dimensions. This lack of regulatory clarity was also noticeable in the bibliometric analysis, where few studies explicitly discussed governance structures or ethical oversight mechanisms. Thus, our findings provided empirical support to highlight the regulatory and assessment gaps that are still present in the field.

Furthermore, the issue of limited accessibility, especially for small and medium-sized enterprises, was derived from the geographical distribution of the studies analyzed. Our results showed a concentration of scientific production in technologically advanced countries, which implicitly reflects barriers to AI adoption in developing regions and smaller organizations. This observation is not only aligned with existing literature but was reinforced by the specific trends we identified in our dataset, where contributions from less developed countries were scarce. Therefore, the formulation of this gap is a contribution that integrates both our empirical findings and the broader research context.

Additionally, the emphasis on the need for interdisciplinary approaches and improved data quality emerged from the analysis of the most frequently used keywords and methodologies. The predominance of technological terms, with limited integration of environmental or social perspectives, revealed a siloed approach in current studies. Our review highlighted that many papers focus on technical efficiency without deeply addressing sustainability’s social or systemic dimensions. As such, the identified gaps do not merely reflect the literature but are strongly grounded in the specific trends and deficiencies detected through our systematic analysis.

4.4. Research Agenda

The development of artificial intelligence in the sustainable economy necessitates a research approach that integrates emerging trends with existing gaps in the literature. The intersection of artificial intelligence (AI) and sustainability has given rise to novel concepts and applications, though there are still lacunae in their implementation and evaluation.

A fundamental pillar of this initiative is the AI-Based Circular Economy, a concept that necessitates the development of more precise models to optimize the utilization of resources and minimize waste. The absence of a universally accepted set of methodologies has hindered its implementation, necessitating research into effective algorithms and integration strategies across diverse industrial sectors.

The advancement of Smart Cities represents a further priority line of study. The utilization of artificial intelligence (AI) has the potential to enhance various aspects of environmental sustainability, including improved mobility, reduced energy consumption, and optimized waste management. However, the absence of explicit metrics poses a significant challenge in accurately assessing the true impact of AI on these environmental objectives. The development of analytical frameworks that quantify environmental and economic benefits is imperative to facilitate implementation in diverse urban environments.

The relationship between AI and the Sustainable Development Goals (SDGs) also requires further exploration. Despite the fact that these objectives inform global policies, the absence of AI-based tools for monitoring them hinders their effective implementation. The necessity to design intelligent systems that analyze data in real time and optimize resource allocation for sustainability is paramount.

From a methodological perspective, research must integrate quantitative and qualitative approaches. The assessment of the impact of AI on sustainability is dependent on two key factors: the first is the use of big data and machine learning, and the second is the incorporation of perception studies that analyze the adoption of these technologies by companies and governments. Moreover, the development of case studies will facilitate the identification of replicable models and best practices.

A further critical gap pertains to the ethical and regulatory implications of AI in sustainability. Automation has the potential to engender disparities in the employment sector and exert an influence on social equity, a phenomenon that necessitates the implementation of regulatory frameworks designed to mitigate these consequences. It is imperative that research is conducted into governance mechanisms that can balance the pursuit of technological development with the upholding of ethical and sustainable principles.

The advancement of AI in the field of sustainability has given rise to a number of unresolved challenges. It is imperative that future research addresses the identified gaps to ensure that the adoption of these technologies contributes effectively to the sustainable transformation of the economy and society.

Future research should critically explore the potential negative externalities associated with AI-driven efficiency in sustainable models. While artificial intelligence offers significant opportunities for optimizing resource use and reducing environmental impact, it may also generate unintended rebound effects. For example, improvements in process efficiency can lead to increased consumption if not accompanied by appropriate regulatory or behavioral frameworks. Additionally, the extensive deployment of AI technologies raises concerns regarding data privacy, as the collection and processing of large volumes of information can potentially compromise individual rights and security.

Moreover, it is essential to investigate the social inequalities that may arise from the accelerated adoption of AI, particularly in the context of labor markets. Automation and digitalization processes can displace traditional jobs, disproportionately affecting low-skilled workers and exacerbating social and economic disparities. Future studies should consider the development of inclusive frameworks that address these social challenges, ensuring that the benefits of AI-driven sustainability are equitably distributed across different populations and regions. Integrating these critical perspectives will contribute to a more holistic and ethically responsible approach to AI in the sustainable economy.

4.5. Implications

The utilization of artificial intelligence within the sustainable economy has the potential to reconceptualize the existing frameworks of innovation, production, and consumption. This study makes a significant contribution to the extant literature by underscoring the necessity for interdisciplinary approaches that seamlessly integrate artificial intelligence with sustainability principles. Despite the extensive coverage of technical applications in the extant literature, there remains a paucity of theoretical models that elucidate the interaction between AI, the circular economy, and sustainable development. It is recommended that future research efforts focus on the development of analytical frameworks capable of assessing the impact of the subject on the structural transformation of production systems and the generation of sustainable value.

In the regulatory sphere, the accelerated integration of AI into economic and environmental processes exposes regulatory gaps that have the potential to generate inequalities and systemic risks. The dearth of specific policies on the responsible use of AI in sustainability hinders the implementation of effective strategies to maximize its benefits and mitigate negative impacts. The establishment of regulatory frameworks that delineate its role in environmental decision-making, natural resource management, and smart city governance is imperative. In the absence of clear regulations, the adoption of AI in this field has the potential to exacerbate socioeconomic disparities, engender biases in resource allocation, and reinforce extractivist models under the guise of algorithmic efficiency.

From a pragmatic standpoint, the integration of AI into the sustainable economy signifies both an opportunity and a challenge for the industry. While the technology has been demonstrated to optimize production processes and reduce environmental impacts, there are a number of technological, economic, and cultural barriers to its implementation. It is incumbent upon companies to develop strategies that combine innovation with social responsibility, thereby preventing digitalization and automation from displacing sustainable practices or affecting labor equity. The absence of standardized metrics to quantify the genuine impact of AI on sustainability is a significant hindrance to its effective implementation. In the absence of rigorous measurement tools, business decisions may prioritize immediate economic criteria, potentially at the expense of long-term objectives aligned with sustainable development.

The advancement of AI in sustainability is contingent on effective coordination between academia, industry, and government to consolidate an innovation ecosystem that ensures equitable benefits for all. It is imperative that future research adopts integrative methodologies that overcome theoretical, regulatory, and practical gaps. In this way, it will be possible to ensure that AI not only transforms the economy, but also does so under ethical and sustainable principles.

4.6. Ethical and Social Risks of AI Use in Strategic Sectors

The application of AI in sectors such as agriculture and mining has demonstrated considerable potential to enhance efficiency and sustainability. However, it is imperative to acknowledge that the implementation of AI also poses significant ethical and social risks. In the domain of precision agriculture, artificial intelligence (AI)-driven tools have the potential to optimize resource utilization and enhance crop yields. However, these systems frequently depend on data models that may exclude local agricultural practices, thereby exacerbating digital divides and marginalizing small-scale farmers. Furthermore, the integration of automation in mining operations, including predictive maintenance and ore grade control, has the potential to displace human labor and exacerbate socio-economic vulnerabilities in already fragile communities.

Recent bibliometric analyses indicate that a significant proportion of the current literature focuses on the technological promise of AI within the context of sustainability, yet pays limited attention to the societal consequences of its deployment [30,32]. While AI is often presented as a primary catalyst for smart and sustainable transitions, particularly in smart cities and industrial operations, it is imperative to consider how disparities in access to technology and data infrastructures may exacerbate existing inequalities. These concerns are especially salient in the Global South, where the absence of robust digital ecosystems can result in disparate implementation and diminished local benefits.

In order to ensure that AI truly contributes to sustainable development, it is necessary to foster inclusive models that address ethical risks and promote community-centered innovation. As [31] have noted, the integration of AI into sustainable systems must extend beyond mere technical feasibility to encompass frameworks that evaluate social impact, equity, and accountability. Collaborative governance models, stakeholder engagement, and context-sensitive design have been identified as mechanisms that can help mitigate the risks of algorithmic bias, environmental externalities, and community exclusion. These mechanisms ensure that artificial intelligence (AI) serves as a tool for just and inclusive development rather than a source of further marginalization.

4.7. Limitations

The study presents methodological and scope limitations that affect the interpretation of the results. The bibliometric approach has the disadvantage of preventing the analysis of grey literature, including conference proceedings, technical reports, and other non-peer-reviewed sources, which may offer valuable insights, especially in fast-evolving fields such as artificial intelligence. The selection of sources may be subject to bias, as certain relevant studies may not be indexed in the consulted databases. The absence of standardized metrics engenders challenges in evaluating the impact of AI on sustainability and restricts the generalizability of results. It is recommended that future studies embrace qualitative approaches and case studies in order to facilitate a more profound exploration of the application of AI in various sectors and thereby enhance the validity of the findings.

A significant limitation of this study lies in the search strategy employed in the Scopus and Web of Science databases, which was restricted exclusively to the TITLE and TS fields. Although this approach ensures a certain level of rigor, it may have limited the comprehensiveness of the review by not including synonyms or related terms associated with artificial intelligence and the sustainable economy, such as “smart sustainability,” “green economy,” or “digital transformation.” The selection of search terms and the decision to exclude other fields, such as abstracts or keywords, may have led to the omission of relevant studies, potentially affecting the representativeness of the results. Future research should consider adopting broader search strategies and provide a detailed justification for the selection of search terms to enhance coverage and minimize the risk of missing significant literature.

A further limitation of the present study is the relatively small number of articles included in the final analysis. This phenomenon can be attributed to two primary factors: first, the rigorous inclusion criteria employed in the study, and second, the nascent stage of development of the topic within academic literature. While this limitation restricts the scope of the review, it underscores the necessity for further exploration and corroborates the importance of promoting research on AI applications within sustainable economic models.

5. Conclusions

A thorough analysis of scientific publications pertaining to artificial intelligence and the sustainable economy reveals a consistent growth trajectory since 2019, signifying an escalating level of interest and the establishment of this subject as a prominent field of study. Additionally, there has been a transition from exploratory research to a more structured empirical approach. The exponential increase in publications indicates the need to understand the relationship between AI and sustainability in different productive sectors, such as manufacturing, waste management, and sustainable digitalization. Conversely, the geographical distribution of studies indicates a concentration in countries with advanced technological development and institutional capacities, resulting in disparities in knowledge access in regions with diminished participation. Italy’s research output exemplifies this, with its findings underscoring the nexus between digital transformation and sustainability in industrial and business contexts.

The identification of the primary authors and journals enables the mapping of the academic community’s structure in this domain. It is evident that there are researchers who demonstrate high productivity yet exhibit low citation impact. This underscores the necessity to enhance the dissemination and utilization of these studies within real-life contexts. The existence of journals with a high citation impact serves to confirm their role as reference channels in scientific discussion.

The analysis of emerging and declining keywords demonstrates the evolution of approaches in this field. Concepts such as the AI-based circular economy, the Internet of Things (IoT), and smart cities have gained relevance, while others have lost relevance. In order to address the challenges that are emerging in this field and to optimize the impact of AI on sustainability, it is necessary to adjust research methodologies. The consolidation of standardized metrics and empirical studies will facilitate the formulation of theoretical models that explain the interaction between AI and sustainable development, ensuring a positive and equitable impact on the global economy. A comprehensive and coherent framework has been developed to illuminate the lines of research, the benefits, and the challenges associated with the utilization of AI for the purpose of achieving a sustainable economy.

6. Recommendations

Based on the research conducted thus far, it is recommended that inter- and transdisciplinary research be promoted. This research should articulate diverse perspectives from the engineering, economic, environmental, and social fields in order to address sustainability in its three broad spheres. This will facilitate a comprehensive approach to the actions undertaken to address the impact of AI on sustainability and to enhance the ethical, technical, and social implications of its use.

In order to enhance research capacities in areas characterized by a paucity of academic output, it is recommended that mechanisms for transnational academic cooperation and collaboration be established. These mechanisms would facilitate the decentralization of research on the subject exclusively to developed countries. Furthermore, it is imperative to devise programs that facilitate the integration of AI not only in large companies but also in SMEs and, in general, throughout the entire productive system. This will contribute to the democratization of these processes and the improvement of emerging economies.

In addressing the ramifications of AI on the sustainable economy, regulatory oversight is imperative, encompassing both technical and ethical dimensions. This regulatory framework is pivotal in promoting the mitigation of the environmental impact arising from the utilization of digital technologies. This assertion is substantiated by research in emerging domains such as AI-Based Circular Economy, Smart Cities, Green IoT, and energy efficiency. Finally, it is imperative to establish systems for the assessment and measurement of the environmental impact of the generation of AI-based technologies (energy consumption of digital infrastructures, cloud computing, and edge computing).