Generative artificial intelligence, or AI, has become a common tool in daily life, helping people with tasks such as writing emails, summarizing documents, translating text, creating images, and even planning vacations. However, this convenience relies on complex systems, including servers, data centers, networks, cooling systems, and computer equipment. These infrastructures are essential for running AI applications but also have significant environmental and resource implications. In July 2026, the French environmental agency ADEME published a report examining the environmental impact of generative AI. The report emphasized the high energy consumption of data centers, the resources required to manufacture equipment, the use of water for cooling, and the conversion of natural land into artificial spaces for technology. In France, much of the digital activity depends on data centers located abroad, many of which rely on electricity that is more carbon-intensive than domestic sources. The challenge is not necessarily to avoid AI but to use it more thoughtfully. For simple tasks such as correcting minor errors in a sentence or performing basic calculations, AI may not be the most efficient or appropriate tool. Before using an AI assistant, it is often useful to ask: does this task really require artificial intelligence? This reflection can help avoid over-reliance on AI for tasks that could be handled more simply. As with many digital tools, the goal is not to eliminate them but to use them intentionally. A clear and specific request usually leads to a faster and more accurate response. On the other hand, repeatedly asking for variations or reformulations can lead to unnecessary computational work. Preparing the request, providing relevant context, and specifying the desired format can improve the effectiveness of AI tools. The ease of generating images with AI has made it possible to create multiple variations of an image in seconds. This accessibility might encourage a pattern of generating many images but keeping only a few, which can lead to excessive resource use. When using AI image generators, it is better to think about the desired outcome in advance and avoid unnecessary iterations that do not add value. Saving useful responses—such as summaries, letter templates, or outlines—can be more efficient than generating them repeatedly. This approach allows for the gradual development of a personal resource library rather than constantly relying on new AI tools. AI can also improve efficiency by quickly analyzing large amounts of information, translating content, automating repetitive tasks, or helping to organize resources. The key is to distinguish between useful applications and habitual, reflexive use. AI is now a standard part of our digital world. Learning to use it with moderation can become a form of digital sobriety—reducing unnecessary requests, increasing precision, and using tools only when they provide real value.