As AI workloads continue to grow, their energy footprint has become a critical concern for both researchers and industry professionals. From document engineering to generative tasks, AI models are the backbone of modern processing pipelines, raising the important question of the associated energy costs.

On Tuesday, August 25, 2026, iCoSys is proud to present a dedicated half-day tutorial at the ACM Symposium on Document Engineering (DocEng’26) in Fribourg. Led by Loïc Guibert, Jean Hennebert, and Sébastien Rumley, the session titled “How to Measure the Energy and Environmental Impact of AI Models? Challenges and Practical Solutions” will provide a comprehensive look at auditing and optimizing the environmental footprint of AI.

The tutorial will cover: – Lifecycle Assessment (LCA) frameworks to quantify operational energy and carbon intensity. – The use of open-source telemetry tools to audit AI workflows. – Practical efficiency techniques to minimize impact, such as quantization, distillation, and pruning. – Strategic deployment aligned with new reporting directives, including the EU CSRD.

To bridge the gap between theory and practice, the session will feature a hands-on live demonstration. A physical benchmark server, directly connected to a Power Distribution Unit (PDU), will be brought to the venue. Participants will be able to observe real-time power drainage measurements while AI workloads run live.

Links: – Register for DocEng’26: https://lnkd.in/eQtQj4tK – More info on DocEng’26: https://doceng.org/doceng2026