This study conducts a structured analysis of scientific contributions dealing with artificial intelligence (AI) and sustainability. By applying the JS-LDA (JavaScript Latent Dirichlet Allocation) thematic modeling to a dataset of 1295 research articles obtained from Web of Science using the narrowly focused query “artificial intelligence AND sustainab*” in the titles or author keywords of the contributions, this work provides a structured landscape of research on the reciprocal interdependencies between the development of artificial intelligence and the objectives of sustainability and sustainable development, quantifying thematic trends and minimizing subjectivity in thematic classification. The scientific landscape, which has been forced into 15 topics that cover the different aspects of the relationship between sustainability and AI, reveals a fragmentation of research interests, ranging from general topics such as environmental integration and efficiency and economic and application-driven approaches (the topics which scholars pay the most attention) to focused research areas such as corporate adoption of green innovation, education and AI in learning, and others. To help clarify this fragmentation, the correlations between the different topics are analyzed, thus indicating areas of scientific interest shared between disciplinary backgrounds and therefore of potential collaboration and cross-fertilization. From a methodological perspective, the use of unsupervised machine learning techniques in literature review ensures reproducibility and establishes a computational framework that can efficiently process and analyze large and growing datasets without losing accuracy or efficiency. Results of this standardized research framework supports policy formation, academic research and industry decision-making, indicating the ways in which AI applications can align with global sustainability goals, including the United Nations Sustainable Development Goals (SDGs).
Mapping AI-Driven Sustainability Research: A Meta-Review Using Topic Modeling / Iandolo, F., La Sala, A., Maielli, G., Vito, P.. - (2026), pp. 335-351. (Digital Transformation Society International Conference – DTS 2025 Parigi ).
Mapping AI-Driven Sustainability Research: A Meta-Review Using Topic Modeling
Francesca Iandolo;Antonio La Sala;Pietro Vito
2026
Abstract
This study conducts a structured analysis of scientific contributions dealing with artificial intelligence (AI) and sustainability. By applying the JS-LDA (JavaScript Latent Dirichlet Allocation) thematic modeling to a dataset of 1295 research articles obtained from Web of Science using the narrowly focused query “artificial intelligence AND sustainab*” in the titles or author keywords of the contributions, this work provides a structured landscape of research on the reciprocal interdependencies between the development of artificial intelligence and the objectives of sustainability and sustainable development, quantifying thematic trends and minimizing subjectivity in thematic classification. The scientific landscape, which has been forced into 15 topics that cover the different aspects of the relationship between sustainability and AI, reveals a fragmentation of research interests, ranging from general topics such as environmental integration and efficiency and economic and application-driven approaches (the topics which scholars pay the most attention) to focused research areas such as corporate adoption of green innovation, education and AI in learning, and others. To help clarify this fragmentation, the correlations between the different topics are analyzed, thus indicating areas of scientific interest shared between disciplinary backgrounds and therefore of potential collaboration and cross-fertilization. From a methodological perspective, the use of unsupervised machine learning techniques in literature review ensures reproducibility and establishes a computational framework that can efficiently process and analyze large and growing datasets without losing accuracy or efficiency. Results of this standardized research framework supports policy formation, academic research and industry decision-making, indicating the ways in which AI applications can align with global sustainability goals, including the United Nations Sustainable Development Goals (SDGs).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
