A practitioner guide for managing AI resource efficiency, using a five-stage management cycle to determine proportionate attention, establish the management basis, manage demand, assess available evidence and turn it into decisions.
Overview
AI is becoming embedded across everyday organisational work, while the environmental resources behind that use remain difficult for most organisations to see or attribute. This guide approaches the issue as a practical environmental-management challenge: organisations do not need perfect impact data before they can start deciding what level of attention is proportionate and how AI demand should be managed.
The guide sets out a five-stage management cycle covering the level of attention required, scope and accountability, demand management, evidence, and reporting and decision-making. At its operational core is a simple hierarchy: Choose an appropriate solution, Reduce unnecessary activity, and Optimise justified use. It also distinguishes between management evidence, computational indicators and environmental information, recognising that each answers a different question and has different limitations.
The aim is not to turn environmental practitioners into AI engineers or to provide a method for calculating an organisation’s AI footprint. Instead, it shows how familiar environmental-management principles can be applied while measurement remains incomplete: establish accountability, challenge unnecessary demand, understand what available metrics can and cannot tell you, and bring evidence back into management so the response can improve as AI use and the information landscape evolve.