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

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article presents a systematic literature review based on the PRISMA 2020 methodology, with the objective of mapping emerging trends in the use of artificial intelligence to promote a sustainable and circular economy. The research identifies key authors, journals, thematic areas, and gaps in the scientific literature on the topic, proposing a research agenda aligned with contemporary challenges.

Positive Aspects

• Relevance and Timeliness: The topic is pertinent in light of the increasing digitalization of production processes and global environmental urgencies.
• Scientific Contribution: The article offers an original contribution by systematizing the state of the art and proposing directions for future investigations.
• Methodological Rigor: The use of the PRISMA 2020 methodology, combined with bibliometric analysis based on consolidated databases (Scopus and Web of Science), ensures the reliability of the study.
• Organization and Clarity: The manuscript is well structured, with clearly defined sections and precise academic writing.

Suggestions for Improvement

Despite the overall consistency of the article, I recommend the following minor revisions to enhance its clarity and impact:

• Critical Regional Discussion: It is suggested to deepen the analysis regarding the scarcity of scientific production in countries of the Global South, addressing epistemic inequalities.
• Ethical Exploration: Expand the section on ethical implications by providing concrete examples of the risks and dilemmas related to the application of AI in sensitive socio-environmental contexts.

Conclusion and Recommendation

This is a relevant, well-founded article that provides significant contributions to the field of digital sustainability. The recommendations suggested are minor and do not compromise the overall quality of the work.

I recommend the acceptance of the manuscript.

Author Response

We thank the editorial team and the reviewer for their significant intellectual contributions to the article during this review process. Below, we list the changes made point by point:

Comment: Critical Regional Discussion: It is suggested to deepen the analysis regarding the scarcity of scientific production in countries of the Global South, addressing epistemic inequalities.

Response: Two paragraphs were generated with examples from countries with low and high scientific production (China, Canada, the US, and Spain). North-South collaborations were proposed as a way to foster capacity and scientific equity.

Comment: Ethical Exploration: Expand the section on ethical implications by providing concrete examples of the risks and dilemmas related to the application of AI in sensitive socio-environmental contexts.

Response: A subsection was added addressing the ethical and social risks of using AI in strategic sectors such as agriculture and mining. Furthermore, the analysis was supported by recent bibliometric studies that warn of inequalities and biases in its implementation.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This paper investigates the role of Artificial Intelligence in promoting a sustainable economy, focusing on resource optimization and environmental impact mitigation. While the topic is relevant and timely, several areas require clarification and refinement to enhance the paper’s academic rigor and overall readability.

1. The manuscript includes numerous abbreviations that are not defined. For example, “AI” (Line 42) is used without being introduced earlier. Please ensure that all abbreviations are defined upon first use, particularly in the introduction.There is inconsistent and unnecessary use of capital letters throughout the manuscript. For instance, “Artificial Intelligence” is capitalized in Line 38, while “artificial intelligence” appears in lowercase on Line 109. Please review the manuscript to maintain consistency in capitalization, following academic style conventions.
2. The five guiding questions outlined in the introduction lack a clear explanation of how they were derived. What criteria were used to formulate them? Does the term “main author” refer to the most frequently cited author, or is it based on publication volume or another metric? Clarifying the methodology used to identify these questions and authors will improve the transparency and reliability of the research.
3. The manuscript references data from 2025, a year that is still ongoing. Please specify up to which month the data were collected. Additionally, explain how data from partial-year 2025 are normalized or compared with previous years in Figure 2. This is crucial for ensuring the validity and comparability of the analysis.
4. The total number of retrieved articles (fewer than 100) appears to be relatively small, raising questions about the comprehensiveness and representativeness of the literature review. Please justify the selection criteria, search strategy, and keyword choice. Moreover, consider discussing whether this research topic is emerging or niche, and how that might impact the volume of available literature.
5. Figure 5 presents country-based article counts, but the methodology for attributing papers to countries is unclear. How are co-authored articles with contributors from multiple countries handled? Clarify whether such articles are counted once for each country or assigned based on first authorship or affiliation distribution.
6. The discussion of other scholars’ work in Section 4.1 contains inconsistent verb tenses. Please revise this section using a consistent past tenseto maintain academic style.

Author Response

We thank the editorial team and the reviewer for their significant intellectual contributions to the article during this review process. Below, we list the changes made point by point:

 

Comment: 1. The manuscript includes numerous abbreviations that are not defined. For example, “AI” (Line 42) is used without being introduced earlier. Please ensure that all abbreviations are defined upon first use, particularly in the introduction.There is inconsistent and unnecessary use of capital letters throughout the manuscript. For instance, “Artificial Intelligence” is capitalized in Line 38, while “artificial intelligence” appears in lowercase on Line 109. Please review the manuscript to maintain consistency in capitalization, following academic style conventions.

Response: A table was included after the keywords defining the acronyms.

Comment: 2. The five guiding questions outlined in the introduction lack a clear explanation of how they were derived. What criteria were used to formulate them? Does the term “main author” refer to the most frequently cited author, or is it based on publication volume or another metric? Clarifying the methodology used to identify these questions and authors will improve the transparency and reliability of the research.

Response: A paragraph was included in the introduction justifying the questions posed.

 

Comment: 3. The manuscript references data from 2025, a year that is still ongoing. Please specify up to which month the data were collected. Additionally, explain how data from partial-year 2025 are normalized or compared with previous years in Figure 2. This is crucial for ensuring the validity and comparability of the analysis.

Response: It was specified that the 2025 data were collected through May 15, and a linear projection was applied to normalize them.

Comment: 4. The total number of retrieved articles (fewer than 100) appears to be relatively small, raising questions about the comprehensiveness and representativeness of the literature review. Please justify the selection criteria, search strategy, and keyword choice. Moreover, consider discussing whether this research topic is emerging or niche, and how that might impact the volume of available literature.

Response: The selection process section was expanded, and a limitations paragraph was created.

Comment: 5. Figure 5 presents country-based article counts, but the methodology for attributing papers to countries is unclear. How are co-authored articles with contributors from multiple countries handled? Clarify whether such articles are counted once for each country or assigned based on first authorship or affiliation distribution.

Response: It was clarified that country allocation was based on the primary affiliation of the first author, according to Scopus and Web of Science. In cases of co-authorship, articles were not duplicated between countries.

Comment: 6. The discussion of other scholars’ work in Section 4.1 contains inconsistent verb tenses. Please revise this section using a consistent past tenseto maintain academic style.

Response: The wording of section 4.1 was reviewed and corrected.

 

Reviewer 3 Report

Comments and Suggestions for Authors

1. Scopus and Web of Science search strategy equations employ just the TITLE or TS fields and intersect AI-related terms with "Circular Economy" or "sustainable economy" without excluding related synonyms (for example, "smart sustainability," "green economy," "digital transformation"). Employing a restrictive search strategy can lead to overlooking pertinent studies. Authors should justify their selection of search terms and consider the possible effect of this limitation on the review's comprehensiveness.

2. Table 1 shows the top ten most highly cited articles but doesn't give us data on why they are most influential (e.g. rigorous methodology, new frameworks). The authors need to examine the content of these articles and discuss how their methods or findings add something new to the literature.

3. The paper describes exponential growth in publications from 2019-2025 and calculates R² as 0.8588 but does not describe how future publications in the year 2025 were "projected." Authors need to clarify if projections are made on the basis of linear extrapolation, exponential forecasting, or otherwise, and justify the use of such projections in bibliometric studies.

4. Figure 6 presents keywords plotted into quadrants but not those criteria (e.g. threshold frequencies, co-occurrence cutoffs) used in labeling keywords "emerging," "growing," or "declining." The authors need to specify those criteria and also present an argument for their methodological decisions.

5. The authors mention the use of Microsoft Excel in data processing and analysis. Excel won't do for heavy network analysis or sophisticated clustering, considering the complexity of bibliometric analysis. The authors need to explain why they used Excel instead of more advanced bibliometric packages such as VOSviewer, Bibliometrix, or Gephi, and detail any limitations.

6. The risk of bias section recognizes the risk of exclusion of non-indexed literature but fails to mention how grey literature or conference papers (which can be very relevant in rapidly developing AI subjects) were treated. Authors need to clarify this aspect and provide an explanation for inclusion/exclusion criteria for grey literature.

7. The three-class classification of authors and journals by impact is presented by the authors without stating the thresholds or algorithms employed in deciding these classes. Authors need to specify whether or not quartiles, cluster analysis, or another approach was employed, and why.

8. Table 2 presents gaps and questions in research but does not connect these gaps to the findings of their own analysis. Authors need to clarify how their own findings contributed to the formulation of these gaps, and whether these represent novel insights or simply a mirror of existing literature.

9. We are informed in the discussion that Fraga-Lamas et al. [13] and Bag et al. [12] are very influential, yet there is no critical analysis of why these papers have been so influential. The authors need to conduct a citation context analysis (e.g. thematic mapping) to discuss the nature of their influence.

10. The authors repeatedly say that AI-driven efficiency helps with sustainability but fail to critically analyze possible negative externalities like rebound effects, data privacy concerns, and social inequality caused by job losses.

11. The limitations section mentions that the bibliometric approach rules out investigation of empirical applications. However, as some of the most highly cited papers identified (e.g. Bag et al. [12]) are empirical, the authors must clarify why case studies or empirical verification of their identified trends were not presented or discussed, and propose the carrying out of future empirical research as a research agenda.

Author Response

We thank the editorial team and the reviewer for their significant intellectual contributions to the article during this review process. Below, we list the changes made point by point:

 

Comment: 1. Scopus and Web of Science search strategy equations employ just the TITLE or TS fields and intersect AI-related terms with "Circular Economy" or "sustainable economy" without excluding related synonyms (for example, "smart sustainability," "green economy," "digital transformation"). Employing a restrictive search strategy can lead to overlooking pertinent studies. Authors should justify their selection of search terms and consider the possible effect of this limitation on the review's comprehensiveness.

Response: This restriction was mentioned in the study limitations.

Comment: 2. Table 1 shows the top ten most highly cited articles but doesn't give us data on why they are most influential (e.g. rigorous methodology, new frameworks). The authors need to examine the content of these articles and discuss how their methods or findings add something new to the literature.

Response: The analysis of the results was expanded with respect to Table 1.

Comment: 3. The paper describes exponential growth in publications from 2019-2025 and calculates R² as 0.8588 but does not describe how future publications in the year 2025 were "projected." Authors need to clarify if projections are made on the basis of linear extrapolation, exponential forecasting, or otherwise, and justify the use of such projections in bibliometric studies.

Response: The paragraph before Figure 2 was corrected to include linear growth.

Comment: 4. Figure 6 presents keywords plotted into quadrants but not those criteria (e.g. threshold frequencies, co-occurrence cutoffs) used in labeling keywords "emerging," "growing," or "declining." The authors need to specify those criteria and also present an argument for their methodological decisions.

Response: This categorization depends on the cutoff point defined for the four quadrants. Specifically, the frequency threshold was set at 50% of the maximum observed frequency, and temporal classification was based on the average occurrence within the last five years. These parameters guided the identification of emerging, growing, and declining keywords and directly determined the location of the terms within each quadrant.

 

Comment: 5. The authors mention the use of Microsoft Excel in data processing and analysis. Excel won't do for heavy network analysis or sophisticated clustering, considering the complexity of bibliometric analysis. The authors need to explain why they used Excel instead of more advanced bibliometric packages such as VOSviewer, Bibliometrix, or Gephi, and detail any limitations.

Response: The present study, in its research questions, does not consider clustering processes. Therefore, the analyses performed on years, authors, journals, countries, and keywords, which are primarily based on frequencies and basic descriptive statistics, can be adequately developed without the need to apply advanced bibliometric analysis techniques within the tool used.

Comment: 6. The risk of bias section recognizes the risk of exclusion of non-indexed literature but fails to mention how grey literature or conference papers (which can be very relevant in rapidly developing AI subjects) were treated. Authors need to clarify this aspect and provide an explanation for inclusion/exclusion criteria for grey literature.

Response: It was clarified that gray literature, including conference papers, was omitted due to the lack of peer review, prioritizing studies that guaranteed methodological rigor and scientific quality.

 

Comment: 7. The three-class classification of authors and journals by impact is presented by the authors without stating the thresholds or algorithms employed in deciding these classes. Authors need to specify whether or not quartiles, cluster analysis, or another approach was employed, and why.

Response: The classification into three groups of authors and journals was performed descriptively, based on observation of the data and their relative frequencies, without the use of formal algorithms, quartiles, or cluster analysis. This grouping was intended to facilitate visual interpretation and highlight evident differences in productivity and citation impact, rather than applying strict statistical segmentation. For future studies, it is recommended to deepen this analysis using more advanced clustering or segmentation techniques that allow for the establishment of more rigorous and objective thresholds.

Comment: 8. Table 2 presents gaps and questions in research but does not connect these gaps to the findings of their own analysis. Authors need to clarify how their own findings contributed to the formulation of these gaps, and whether these represent novel insights or simply a mirror of existing literature.

Response: It was clarified that the gaps identified in Table 2 are directly linked to the analysis' findings, especially in aspects such as the lack of regulatory frameworks, the geographic concentration of studies, and the limited focus on interdisciplinary perspectives, reinforcing that these gaps arise from both the literature and the results obtained in this study.

Comment: 9. We are informed in the discussion that Fraga-Lamas et al. [13] and Bag et al. [12] are very influential, yet there is no critical analysis of why these papers have been so influential. The authors need to conduct a citation context analysis (e.g. thematic mapping) to discuss the nature of their influence.

Response: The discussion was expanded to include a critical analysis explaining the relevance of the studies by Bag et al. [12] and Fraga-Lamas et al. [13], highlighting their impact due to the solidity of their theoretical frameworks, the integration of novel approaches, and their broad application in subsequent research on sustainability and digital transformation.

Comment: 10. The authors repeatedly say that AI-driven efficiency helps with sustainability but fail to critically analyze possible negative externalities like rebound effects, data privacy concerns, and social inequality caused by job losses.

Response: The need to critically analyze the potential negative externalities of AI, including rebound effects, data privacy risks, and the social impact of job losses, was added to the research agenda, proposing these topics as future lines of study.

Comment: 11. The limitations section mentions that the bibliometric approach rules out investigation of empirical applications. However, as some of the most highly cited papers identified (e.g. Bag et al. [12]) are empirical, the authors must clarify why case studies or empirical verification of their identified trends were not presented or discussed, and propose the carrying out of future empirical research as a research agenda.

Response: The limitations section was adjusted, replacing the reference to empirical applications with the exclusion of gray literature, such as non-peer-reviewed papers and reports, recognizing its relevance in rapidly developing fields such as artificial intelligence.

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The paper can be accepted now.

Reviewer 3 Report

Comments and Suggestions for Authors

Accepted