Your AI habit is wasting precious resources. Here’s how to use it responsibly

Mukesh Sharma/Unsplash If someone used a large truck to deliver one envelope across the street, what would your reaction be? You would probably say it worked, but it was wasteful. The envelope arrived, but the method made little sense. In many ways, this mirrors how we often use artificial intelligence (AI) today. We use powerful AI systems such as ChatGPT to write short messages, polish simple sentences, and answer questions we could handle ourselves with less intensive tools. The outcome is the same. But under the hood, the differences in processing, electricity, and water use are enormous. One small request will not change the planet. But millions of small and unnecessary requests can add up over time. The question is not whether AI is good or bad. The better question is whether we are using the right amount of AI for the right task. AI is becoming a utility AI is quickly becoming an everyday utility, like electricity. But because it feels invisible, we can easily forget every use has a cost . We know not to leave lights on all day. We also know not to run an air conditioner with windows open. We need a similar mindset for AI. AI is often described as living in “the cloud”. This makes it sound light, clean, and almost magical. But the cloud is not really a cloud. It is someone else’s computer. More precisely, it is a large network of data centres filled with servers , chips, cables and cooling systems. When we interact with AI systems, they do not simply “know” the answers. They run calculations through large computer systems to generate a response. Training large AI models require enormous resources, but daily use adds up too. Data centres can create significant carbon emissions . They also take up large tracts of land, put pressure on electricity grids and water supplies, and generate electronic waste. The International Energy Agency has projected global electricity use by data centres could roughly double to around 945 terawatt-hours (TWh) by 2030. How to use AI more sensibly This makes it important to consider how we can use AI more sensibly – and sustainably. First, choose the right tool for the task. Not every job needs the most powerful AI model. Sometimes a smaller tool, search engine, calculator, or even your own brain is enough. Ask yourself: is AI adding any value here, or am I using it out of habit? Second, write clearer prompts. Vague questions often lead to many follow-up prompts. A clear first request can reduce this back and forth. It is like giving a taxi driver the correct address before the trip starts instead of constantly correcting the route along the way. Third, ask only for what you need. If you need a short list, do not ask for a long report. More words usually mean more computing. Fourth, be more careful with images, audio, video and other media files. These typically require more computing power than texts. The role of organisations and governments Large organisations have an even bigger responsibility. They should not add AI to every product just because it is fashionable. Before adopting AI, they should ask: what problem are we solving? Is AI really needed? If it is, can we use a smaller model? Governments should also require large data centres to report electricity use, water use, emissions and electronic waste as part of planning approvals. Australia has released guidelines for data centre developers , but they are voluntary, not law . New data centres should be planned very carefully, especially in places with limited water or stressed grids. Sustainability labels, like energy ratings on appliances, could help users compare tools by efficiency. These measures could make sustainability part of normal AI governance, not an afterthought. They could help ensure AI stops being the equivalent of the truck carrying one envelope across the street. Seyedali Mirjalili does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.