Every time you type a prompt into ChatGPT or generate an AI image, you're consuming water and electricity in ways most users never consider. A new analysis from the University of New Brunswick reveals the staggering environmental cost of AI use at scale, and offers practical steps individuals can take to reduce their digital footprint without abandoning the technology entirely.

The numbers are sobering. With approximately 2.5 billion prompts entered into ChatGPT daily, the cumulative resource demand is enormous. An average ChatGPT text query consumes roughly 200 times more energy than traditional text classification tasks like spam filtering. AI image generation is far worse, requiring 1,450 times more energy than basic text queries. Video generation stands as the most intensive, consuming enough energy to power 200,000 text classifications.

To put this in tangible terms, generating a single AI image carries a water footprint of approximately 29 milliliters, equivalent to two tablespoons. A complex AI-generated video uses around 4.1 liters of water, the same amount a person would drink over two days.

How Massive Is AI's Global Energy Footprint?

When scaled globally, the environmental impact becomes staggering. In 2025 alone, data centers worldwide consumed an estimated 448 terawatt-hours (TWh) of electricity. To contextualize this figure, the United Nations University notes that 448 TWh represents enough electricity to power the residential needs of 1.3 billion people in Sub-Saharan Africa for 2.6 years.

Water consumption tells a similarly alarming story. Global data centers used 4.5 trillion liters of water in 2025, sufficient to meet the annual basic domestic water needs of over 600 million people in Sub-Saharan Africa. According to research cited by the University of New Brunswick, if data center electricity consumption were ranked as a country, it would rank 11th globally in total electricity use.

What Is the "Green AI" Movement Trying to Accomplish?

In response to these environmental concerns, a movement called "Green AI" has emerged to reduce the ecological impact of artificial intelligence development and deployment. This approach focuses on three main strategies: redesigning algorithms and system architectures to be more efficient, adopting more efficient hardware and infrastructure, and transitioning data centers to renewable energy sources. These initiatives are designed to support the United Nations Sustainable Development Goals (SDGs).

However, experts emphasize that individual users also have a role to play. While most people don't participate in the behind-the-scenes development of AI systems, everyone using these tools can make conscious choices about how frequently and intensively they rely on them.

How to Reduce Your AI Environmental Impact

The University of New Brunswick recommends several practical strategies for more sustainable AI use:

• Evaluate necessity: Ask yourself whether a traditional search engine can provide the information you need before turning to AI tools. Don't make AI your automatic first choice.
• Reduce prompt verbosity: Use fewer words in your requests. Remove unnecessary politeness markers like "please" and "thank you," which add to processing demands without adding value.
• Consolidate requests: Explain what you're seeking in a single, concise prompt rather than asking multiple follow-up questions that require additional processing.
• Lower generation specifications: If you need to generate images or videos, opt for lower resolution, fewer frames, and fewer processing steps when possible.
• Reuse previous results: Rather than asking AI to regenerate content you've already received, save and reuse previous outputs.

These individual actions may seem minor, but they accumulate at scale. As the University of New Brunswick notes, "Small skips create ripples". The cumulative effect of millions of users making more conscious choices about AI usage could meaningfully reduce the technology's environmental burden.

The irony is not lost on experts: using a technology with significant environmental costs to solve environmental problems requires careful consideration. Rather than viewing AI as a universal solution, the emerging consensus suggests treating it as one tool among many, deployed thoughtfully rather than reflexively.

The conversation around AI and sustainability is still developing, but one thing is clear: as AI becomes more integrated into daily life, the choices users make about how they engage with these tools will matter. Whether you're a student, professional, or casual user, understanding the hidden environmental cost of each prompt is the first step toward more sustainable technology use.