〈Section 2-5) Transparency in the Use of Generative AI〉 provides guidance on how to approach transparency in this context. Sixth, it is necessary to take into account the impacts of advances in generative AI technologies on the environment and labor. AI systems consume large amounts of resources—such as electricity and water—during both training and operation. This is because AI training and deployment require large-scale computation, and as efforts to improve AI performance continue, the volume of training data and the size of model parameters are also increasing. In proportion to this growth, AI’s energy demand is rising rapidly. Because a significant share of current energy supply still relies on high-carbon sources such as coal and natural gas, concerns are growing that the expansion of AI and data centers is exacerbating the climate crisis. In addition, the large quantities of water consumed to cool data centers have, in some cases, led to conflicts with local communities. Even civil society organizations that are not primarily environmental groups cannot ignore these issues if they recognize the urgency of responding to the climate crisis. To be sure, energy consumption in AI training and operation is a structural issue that individual civil society organizations, as users, have limited ability to influence directly. Nevertheless, organizations can choose to use lightweight models that offer similar functionality while consuming less energy, and they can demand that AI providers make such models available. They can also call on AI companies to transparently disclose data on how much energy is used in the development and operation of AI systems. 62 63