Забезпечення сталого використання генеративного штучного інтелекту підприємствами на основі оптимізації ресурсоспоживання

The object of this study is the impact of using generative artificial intelligence (AI) on the resource efficiency of enterprises in the context of sustainable development. The issue relates to the fact that despite the ability of generative AI to optimize management and production processes, its use is accompanied by an increase in electricity consumption (grid, mostly non-renewable) and water, which creates new environmental risks. This paper systematizes existing methods for indirectly estimating energy and water consumption in the process of generative AI model functioning. Based on this, an approach to assessing the ecological footprint of generative AI has been devised, which takes into account four indicators such as query length, complexity, task type, and industry. A special feature of the proposed approach is a combination of quantitative analysis, regression modeling, and query classification to assess resource intensity. An empirical study has shown that high-volume queries (analytical and creative) generate significantly higher resource consumption (2.1–2.3 Wh of electricity and more than 0.8 liters of water) while factual queries create a minimal load (< 0.12 Wh). This difference is explained by the complexity of information processing and the involvement of significant computing power in cloud data centers. To evaluate the feasibility of AI implementation, a sustainability index has been proposed to assess the balance between the efficiency achieved and resources spent. The proposed approach could be used by enterprises under conditions of limited access to energy and water resources, in particular during post-war recovery and implementation of the principles of sustainable development