Can Energy Vampires make organizations more sustainable? The case of LLMs and BPM

The environmental impact Generative AI – and in particular, Large Language Models (LLMs) – is far from negligi- ble. Despite this, LLMs can help companies rethink sustainable practices, especially in small organizations lacking technical and methodological expertise in ecological transition. Our work has a twofold aim: (i) to investigate whether LLMs can effectively support the reengineering of business processes toward greener alternatives, and (ii) to assess whether employing such energy- intensive technologies aligns with sustainability objectives in a meaningful and cost-effective way. To address these questions, we have designed a three-step methodological Framework currently under implementation. The Framework involves the use of LLMs ”as-as-service” for refactoring textual descriptions of business processes. Preliminary results are promising, with GPT-4o and Claude 3.5 demonstrating strong performance on a benchmark of well-known processes provided by Camunda, while the study on the energetic impact of LLMs in organizational settings is still ongoing.