Feasibility study to evaluate and develop a possible labelling system for AI applications

The NADIKI project has developed a tool that visualises the environmental impacts of AI systems both across their entire life cycle and in real time. It combines the assessment of manufacturing costs with the direct measurement of environmental consumption and CO₂ emissions from individual operations or inferences made by AI applications. The tool is a key prerequisite for informed decisions regarding the environmentally sound design, development and use of AI – for example, through green coding or the climate-conscious deployment of new technologies. As part of the project, we examined the accuracy with which environmental impacts can be measured using a server-side tool and the legal framework that must be considered when using such a tool.