Skip to Main Content
Proctor Library
Today's Hours
View All Library Hours

Generative AI Ethics and Ethical Use in Academic Contexts

This is a guide designed to inform students how to use generative AI tools ethically to enhance their educational journey.

The Hidden Footprint: Energy, Water, & Carbon

While generative AI offers incredible potential, it also comes with a significant environmental cost. The advanced computations required to train and run these powerful models demand vast amounts of energy and other resources. As responsible users of technology, it's important to be aware of this impact and consider how we can contribute to more sustainable AI practices.

A four-panel comic titled "AI and the Planet." Panel 1 (Top Left): A data center with smoke clouds above it. A speech bubble says, “To train big models, you need a lot of computing power!” The panel header reads: "AI Uses a Lot of Energy." Panel 2 (Top Right): A person points to a sad Earth and a smiling smartphone screen. The speech bubble says, “Sorry about that.” The panel header reads: "This Produces Emissions." Panel 3 (Bottom Left): The same person gestures toward solar panels and a wind turbine. The speech bubble says, “With clean energy sources like wind and solar...” The panel header reads: "But We Can Change That." Panel 4 (Bottom Right): The person smiles and gestures toward the phone, which now says “Good call!” The person says, “Let’s be mindful of how much power we really need.” The panel header reads: "And Also by Using AI Responsibly." The comic uses clean, friendly illustrations and emphasizes the environmental impact of AI and the importance of using clean energy and AI responsibly.

Image created by Trina McCowan Adams using ChatGPT- CC BY-NC-SA 4.0

The Hidden Footprint: Energy, Water, & Carbon

Every time you interact with a generative AI model, it relies on powerful computers, typically housed in massive data centers. These data centers consume enormous amounts of electricity, and they also require substantial water for cooling.

  1. Energy Consumption:

    • Training Models: Training a single large language model (LLM) can consume energy equivalent to powering dozens or even hundreds of homes for an entire year. This process involves crunching through vast datasets for weeks or months. For instance, training GPT-3 (a predecessor to current popular models) consumed an estimated 1,287 MWh of electricity – enough to power about 120 average U.S. homes for a year (MIT News, 2025; Adasci, n.d.).

    • Inference (Using the Model): While training is a one-time intensive process, "inference" (when the AI generates a response to your prompt) is a continuous process. Each query uses energy. While a single query's energy use is small, the cumulative effect of millions of users asking billions of questions daily adds up significantly. Some estimates suggest a single ChatGPT query can use 10 times as much energy as a standard Google search (Science News, 2024).

    • Data Centers: These facilities are rapidly becoming major electricity consumers globally, with their energy demands potentially surpassing entire countries (MIT News, 2025).

  2. Water Consumption:

    • Data centers generate immense heat, which requires constant cooling to prevent equipment damage. This cooling often relies on large quantities of water, primarily for evaporative cooling towers.

    • Reports indicate that hyperscale data centers can use millions of liters of water daily. For example, some Microsoft data centers in Iowa used 11.5 million gallons of water in the final months of training GPT-4 (Jisc, 2024). This raises concerns, especially in regions facing water scarcity.

  3. Carbon Emissions:

    • The vast majority of the electricity powering data centers still comes from fossil fuels, leading to significant greenhouse gas emissions. These emissions contribute directly to climate change.

    • Training a single large AI model can generate as much CO2 as several cars over their entire lifespan (The Economic Times, 2025).

Mitigating the Impact: What Can Be Done?

Addressing the environmental footprint of AI requires efforts from developers, companies, and individual users.

What AI Developers & Companies Are Doing (or Can Do):

  • Energy-Efficient Hardware: Developing specialized chips (like neuromorphic chips) and optimizing hardware for lower power consumption.

  • Algorithm Optimization: Creating more efficient AI models and algorithms that can achieve similar performance with less computational power (e.g., model pruning, quantization).

  • Renewable Energy: Shifting data centers to run on 100% renewable energy sources like solar and wind power. Many major tech companies are investing heavily in this.

  • Advanced Cooling Technologies: Implementing more efficient cooling methods, such as liquid immersion cooling, to reduce water consumption.

  • Location Strategy: Placing data centers in colder climates naturally reduces cooling needs.

  • Responsible Scheduling: Scheduling intensive AI computations during off-peak energy hours or when renewable energy sources are abundant on the grid (e.g., sunny or windy days).

What You Can Do (as a User)

While individual queries have a small impact, cumulative actions matter. Consider these practices for "Green AI" use:

  • Be Mindful of Necessity: Before prompting an AI, consider if it's truly necessary. Can a simple search, your own critical thinking, or a traditional tool achieve the same goal with less energy?

  • Be Specific with Prompts: Vague or overly broad prompts might lead to longer generation times and more computational effort. Clear, concise prompts can lead to more efficient responses.

  • Evaluate Value: If the AI's output isn't useful, discard it and refine your approach rather than endlessly generating variations.

  • Support Sustainable Practices: When possible, choose AI tools or services from companies that publicly commit to and report on their environmental sustainability efforts (e.g., those powered by renewable energy, with transparent water usage data).

  • Advocate for Transparency: Encourage AI providers to be more transparent about the energy and water consumption of their models.

Understanding the environmental impact of generative AI is part of being an ethically-minded and responsible digital citizen. By making conscious choices, you contribute to a more sustainable technological future.

References