![[2602.16042] AI-CARE: Carbon-Aware Reporting Evaluation Metric for AI Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16042] AI-CARE: Carbon-Aware Reporting Evaluation Metric for AI Models
Summary
The paper proposes AI-CARE, a carbon-aware evaluation metric for machine learning models, addressing the environmental impact of model training and inference.
Why It Matters
As machine learning grows, its environmental footprint becomes critical. AI-CARE shifts focus from traditional performance metrics to include energy consumption and carbon emissions, promoting sustainable AI practices aligned with global sustainability goals.
Key Takeaways
- AI-CARE introduces a multi-objective evaluation framework for ML models.
- The carbon-performance tradeoff curve visualizes the balance between model performance and carbon emissions.
- Shifting to carbon-aware metrics can change model rankings and encourage eco-friendly architectures.
- The tool aims to align ML advancements with sustainability efforts.
- Empirical validation shows the effectiveness of carbon-aware benchmarking.
Computer Science > Machine Learning arXiv:2602.16042 (cs) [Submitted on 17 Feb 2026] Title:AI-CARE: Carbon-Aware Reporting Evaluation Metric for AI Models Authors:KC Santosh, Srikanth Baride, Rodrigue Rizk View a PDF of the paper titled AI-CARE: Carbon-Aware Reporting Evaluation Metric for AI Models, by KC Santosh and 2 other authors View PDF HTML (experimental) Abstract:As machine learning (ML) continues its rapid expansion, the environmental cost of model training and inference has become a critical societal concern. Existing benchmarks overwhelmingly focus on standard performance metrics such as accuracy, BLEU, or mAP, while largely ignoring energy consumption and carbon emissions. This single-objective evaluation paradigm is increasingly misaligned with the practical requirements of large-scale deployment, particularly in energy-constrained environments such as mobile devices, developing regions, and climate-aware enterprises. In this paper, we propose AI-CARE, an evaluation tool for reporting energy consumption, and carbon emissions of ML models. In addition, we introduce the carbon-performance tradeoff curve, an interpretable tool that visualizes the Pareto frontier between performance and carbon cost. We demonstrate, through theoretical analysis and empirical validation on representative ML workloads, that carbon-aware benchmarking changes the relative ranking of models and encourages architectures that are simultaneously accurate and environmentally responsible. Ou...
