AI can assess energy-conscious enterprises with 90% accuracy

European researchers have developed a pioneering machine learning model that predicts whether companies are managing their energy use sustainably, without needing to install a single sensor. The model, which uses digital maturity indicators to gauge awareness of the energy impact of information technology (IT) and artificial intelligence (AI), is detailed in the paper "Awareness of the Impact of IT/AI on Energy Consumption in Enterprises: A Machine Learning-Based Modelling Towards a Sustainable Digital Transformation", published in Energies.

The study introduces a data-driven approach to measuring energy-conscious behavior in enterprises by analyzing how deeply digital technologies are embedded in their operations. By linking AI intensity, automation levels, analytics adoption, and smart interface integration to sustainability outcomes, the researchers show how AI can be used to assess sustainability itself, replacing traditional audit methods with predictive intelligence.

Can AI accurately predict corporate energy awareness?

Can artificial intelligence identify energy-aware enterprises purely from their digital behavior patterns? To answer this, the team analyzed data from 300 Polish enterprises, spanning manufacturing, services, and IT sectors. Each company was profiled based on five digital maturity indicators:

• Level of AI integration in operations.
• Degree of process automation and algorithmic decision-making.
• Depth of data analytics in management systems.
• Intensity of digital intelligence in core business processes.
• Coverage of smart interfaces across organizational functions.

The researchers then applied five machine learning algorithms, Support Vector Machine (SVM), Logistic Regression (LR), Decision Tree (DT), Neural Network (NN), and k-Nearest Neighbors (KNN), to classify whether each enterprise demonstrated awareness of its IT and AI energy footprint.

The SVM model proved to be the most accurate, achieving 90 percent test accuracy and an F1 score of 89.8 percent, far outperforming other methods. This high precision indicates that corporate sustainability awareness can be effectively inferred from digital transformation patterns alone, even without direct measurements of energy consumption.

Unlike conventional sustainability audits, which require time-consuming data collection and hardware deployment, this AI-based model draws conclusions from existing enterprise data, such as automation levels and digital infrastructure. The result is a fast, low-cost, and non-intrusive way to estimate energy-conscious practices in digitally active companies.

Which digital factors matter most in sustainability awareness?

After training and testing the models, the researchers used Shapley value analysis to determine which features contributed most to predicting sustainable energy management. Two factors stood out:

• Coverage of intelligent interface applications, such as smart dashboards and IoT-enabled controls.
• Intensity of digital intelligence, reflecting how deeply AI and analytics are integrated into daily business operations.

These indicators, the study found, were more decisive than even automation or data analytics alone. Companies that invested in smart interfaces and adaptive AI systems were far more likely to demonstrate active energy awareness and implement sustainable IT policies.

The analysis also showed that digital intelligence amplifies sustainability indirectly by improving process optimization, predictive maintenance, and data-driven resource allocation. In other words, enterprises with higher AI literacy tend to consume energy more efficiently, not through direct control mechanisms but through smarter, more informed decision-making.

The study's framework aligns with the principles of sustainable digital transformation, in which technology adoption is evaluated not just for productivity but also for environmental and ethical impact. By quantifying how digital maturity correlates with sustainability, the authors bridge a critical research gap linking digital transformation, AI governance, and green enterprise policy.

From policy to practice: A new way to benchmark sustainability

The machine learning framework offers a scalable, data-based tool that policymakers, ESG auditors, and business leaders can use to assess sustainability readiness.

The researchers argue that AI-based assessment models could soon complement or even replace traditional sustainability audits, which often rely on self-reported data and physical energy measurements. By contrast, the proposed model uses existing enterprise datasets to infer sustainability awareness, making it particularly valuable for small and medium-sized enterprises (SMEs) that lack the resources for full-scale environmental monitoring.

The system can also be integrated into national energy efficiency programs or ESG reporting platforms, helping governments and corporations quickly identify which sectors or companies are lagging in energy-conscious digitalization. AI can not only improve operational efficiency but also enhance transparency and accountability in how enterprises manage their environmental impact.

To maintain interpretability and policy alignment, the authors advocate for explainable machine learning models, such as SVMs and Decision Trees, over purely black-box neural networks. This approach ensures that decision-makers can understand the rationale behind AI-driven sustainability assessments and use the results for targeted intervention or policy design.

In addition to industry applications, the authors highlight the framework's potential use in education and research, where it can serve as a reference model for developing metrics that link digital maturity to environmental stewardship. The paper's proposed model could also underpin new AI-driven ESG benchmarking systems, capable of continuously monitoring sustainability progress without intrusive data collection.