mini instead of 4o, or Claude Haiku instead of Opus) can be used. Lightweight models require significantly less computational power and energy than large-scale models. For tasks such as simple summarization, organization, translation, or classification, ultralarge models are often unnecessary. To reduce environmental impact, it is important to select models that are appropriate for the specific use case. That said, it may be difficult for users to determine which model is most appropriate in each situation. In this respect, it could be effective for AI providers to develop interfaces that automatically recommend or select suitable models based on the task at hand. In addition, it is important to use products and services from companies that implement environmentally responsible policies, such as disclosing information on the environmental impact assessments of data centers used for AI operations, electricity consumption, and energy efficiency, as well as adopting renewable energy sources. In order to assess which companies are genuinely pursuing such environmentally friendly practices, it is essential that companies first disclose relevant data in a transparent manner. Corporate environmental policies, or ESG reports may serve as useful reference points for this assessment. Generative AI Guide for Civil Society