The rapid advancement and widespread adoption of Artificial Intelligence has brought transformative opportunities across scientific, industrial, and societal domains. At the same time, the growing scale and complexity of AI systems raise critical concerns related to energy consumption, environmental impact, resource usage, and broader societal implications. As modern AI increasingly relies on data-intensive and computation-heavy models, addressing these challenges has become a pressing priority, requiring a shift from purely performance-driven development toward approaches that explicitly account for sustainability, responsibility, and long-term impact.
CEUR Workshop ISSN1613-0073
The workshop continues to serve as a multidisciplinary venue for discussing recent advances in greenaware and sustainability-oriented Artificial Intelligence. This year’s edition further consolidates this vision by structuring the program around three complementary and interconnected thematic sessions, each addressing a key dimension of sustainable AI research and practice.
The first session focuses on Sustainability, broadly understood as the integration of environmental, social, and governance principles into AI system design and analysis. The second session is devoted to Green AI, emphasizing methods, metrics, and tools aimed at reducing the computational, energetic, and environmental costs of AI technologies. Finally, the Applications session highlights concrete deployments of green-aware AI solutions across diverse domains, illustrating how sustainability principles can be efectively translated into practice.
Together, these sessions reflect the workshop’s overarching goal of promoting holistic approaches to AI sustainability, encouraging cross-community dialogue, and advancing solutions that balance performance, responsibility, and long-term impact.
The Sustainability session brings together contributions that frame sustainability in AI as a multifaceted concept encompassing resource eficiency, transparency, governance, and social responsibility. Rather than focusing on isolated metrics, the papers collectively illustrate how sustainability emerges from informed theoretical foundations, practical methodologies, and principled system design.
A first group of contributions addresses resource-aware and eficiency-driven perspectives . The
paper [
A second thematic cluster focuses on transparency, governance, and institutional sustainability. The
paper [
Finally, sustainability is explored from the perspective of responsible and socially grounded AI behavior.
The paper [
Overall, the Sustainability session underscores the importance of integrated approaches that combine eficiency, transparency, and social responsibility, advancing sustainability as a foundational principle for future AI research.
The Green AI session focuses on methodological, algorithmic, and system-level approaches aimed at reducing the environmental footprint of AI systems while preserving efectiveness and reliability. The contributions span model design, learning paradigms, evaluation practices, and incentive mechanisms, ofering a comprehensive view of Green AI across the AI lifecycle.
A first group of papers addresses energy-aware model design and deployment. The paper [
Another key theme concerns data-centric and distributed Green AI. The paper [8] proposes a datacentric framework for environmentally sustainable Federated Learning, leveraging intelligent node selection and data reduction strategies guided by data quality and carbon footprint estimates.
Several contributions emphasize evaluation, accountability, and benchmarking. The paper [9] introduces a systematic framework for measuring energy consumption in AI planning, moving beyond runtime as the sole eficiency proxy. In a complementary direction, [ 10] proposes a methodology for reducing the cost and environmental impact of large language model benchmarking while maintaining reliable performance estimates.
Finally, Green AI is explored through optimization and incentive-aware mechanisms. The paper [11] studies renewable energy auctions using obviously strategy-proof mechanisms, while [12] addresses green-aware decision processes that align algorithmic incentives with energy-eficient behavior.
Collectively, the Green AI session highlights the maturity of the field, demonstrating that meaningful sustainability gains arise from the integration of algorithmic innovation, system design, evaluation methodology, and incentive structures.
The Applications session showcases how green-aware and sustainability-oriented AI techniques can be efectively deployed across diverse real-world domains, demonstrating tangible environmental and societal benefits.
In the context of sustainable manufacturing, [13] presents a deep reinforcement learning approach for optimizing milling processes, explicitly balancing production eficiency and tool lifespan through adaptive decision-making.
In precision agriculture, [14] proposes calibrated semantic segmentation models for drone-based weed mapping. By combining lightweight architectures with post-hoc calibration techniques, the approach enables reliable confidence estimates and more resource-eficient interventions.
The paper [15] addresses wildfire monitoring and management , presenting the PRIMA platform for real-time analysis that integrates heterogeneous data sources with AI-based predictive models and LLM-supported decision interfaces.
Human-centered and socially aware AI systems are explored in [16], which combines symbolic representations with data-driven components to support interpretable and context-aware decisionmaking. Finally, [17] focuses on environmental monitoring, proposing machine learning techniques for extracting actionable insights from environmental data under constrained computational resources.
Overall, the Applications session illustrates the breadth of domains in which green-aware AI can be concretely applied, highlighting how tailored methodologies can efectively address domain-specific sustainability challenges.
The 2nd Workshop on Green-Aware Artificial Intelligence received a total of 23 submissions, nearly tripling the number from the previous edition. Following a rigorous single-blind peer-review process, 17 papers were accepted, corresponding to an acceptance rate of approximately 74%. The final program includes 8 regular papers and 9 short papers.
The accepted contributions reflect the growing convergence between AI research and environmental awareness, addressing ecological, social, and ethical challenges through innovative methods, frameworks, and applications.
Each submission was reviewed by two independent reviewers. Final decisions were made by the workshop chairs based on reviewer feedback and standard academic criteria, including originality, technical quality, relevance to the workshop themes, potential impact on sustainable AI, and clarity of presentation.
We sincerely thank the members of the Program Committee for their dedication and careful evaluations, which were essential in ensuring the quality and rigor of the selected papers. In recognition of their valuable contributions, we gratefully acknowledge the following Program Committee members: • Julien Aligon, Université de Toulouse, France • Vittorio Bilò, University of Salento, Italy • Gianlorenzo D’Angelo, Gran Sasso Science Institute, Italy • Davide Di Stefano, TU Wien, Austria • Diodato Ferraioli, Università di Salerno, Italy • Hong Jia, University of Melbourne, Australia • Martin Lněnička, Univerzita Pardubice, Czech Republic • Cristian Molinaro, University of Calabria, Italy • Gianpiero Monaco, University of Chieti-Pescara, Italy • Luca Moscardelli, University of Chieti-Pescara, Italy • Francesco Mureddu, Lisbon Council, Portugal • Alessio Orsino, University of Calabria, Italy • Domenico Talia, University of Calabria, Italy • Kees Van Berkel, TU Wien, Austria • Bernard Van Gastel, Radboud University, The Netherlands • Cosimo Vinci, University of Salento, Italy
We would like to express our sincere gratitude to everyone who contributed to the success of the 2nd Workshop on Green-Aware Artificial Intelligence . We thank the Program Committee members and reviewers for their dedication, as well as all authors and participants for their valuable contributions and discussions.
A special acknowledgment goes to our invited speaker, Prof. Thomas Eiter (TU Wien), for his insightful talk on The Bilateral AI approach for Green and Sustainable AI. We also thank the organizers of the 28th European Conference on Artificial Intelligence for their support. Finally, we acknowledge the support of the PNRR project FAIR – Future AI Research (PE00000013), Spoke 9 – Green-aware AI, under the NRRP MUR program funded by NextGenerationEU. [7] E. Cruciani, R. Verdecchia, Choosing to be green: Advancing green ai via dynamic model selection, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [8] M. Sabella1, M. Vitali, Eco-friendly ai: a framework for data centric green federated learning, in:
Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [9] I. Georgievski, What is hiding in the energy footprint of ai planning? initiating energy accountability, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [10] N. Lemonnier, J. Kluska, V. Morel, Sustainable llm benchmarking: Eficient evaluation optimizing costs and resources, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [11] D. Ferraioli, Obviously strategy-proof auctions for energy eficiency, in: Proceedings of the 2nd
Workshop on Green-Aware Artificial Intelligence, 2025. [12] S. Capalbo, E. Cesario, P. Lindia, F. Lobello, A. Vinci, Enhancing cloud energy eficiency through predictive machine learning for inter- and intra-data center vm consolidation, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [13] S. Ferrisi, M. Afrasiabi, R. Guido, D. Umbrello, G. Ambrogio, M. Bambach, Deep reinforcement learning-based parameter optimization in milling: A novel approach for enhancing tool life, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [14] P. D. Marinis, G. Detomaso, G. Vessio, G. Castellano, Calibrated weed mapping, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [15] G. Alì, P. A. Fusaro, L. Granata, S. Iiritano, R. Mangiardi, A. Mazza, A. Mirante, P. S. Pantano, The prima project: A real-time integrated platform for forest fire monitoring and analysis, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [16] L. Laboccetta, G. Terracina, F. Calimeri, S. Perri, M. Rufolo, D. Iacopino, M. Maria, S. Iiritano, A human-centric environment (hce) framework for sustainable production in a bakery, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025. [17] V. Camerada, D. Guidotti, S. Lampreu, L. Pandolfo, L. Pulina, Sket-monitor: A knowledge-driven ai system for sustainable environmental and territorial monitoring, in: Proceedings of the 2nd Workshop on Green-Aware Artificial Intelligence, 2025.