As generative artificial intelligence (GenAI) becomes part of daily life, more people are asking: What does this mean for the planet? Maybe you’ve seen headlines about data centers consuming a lot of electricity and water. As society races toward an AI-powered future, it’s natural to wonder: How can we be sure the Earth can sustain this shift?
It's a fair question — and one we take seriously at Pitt Digital. We believe that responsible innovation means considering what technology can do, what it costs in dollars, and what the environmental and social impacts are. The good news? There’s a lot we can do to use AI thoughtfully.
Too Much Power? The Cost of Training vs. Using
Training a generative AI model from scratch requires enormous energy. When you use a Generative AI @ Pitt tool, you're not doing that, you're running queries against pre-trained models, which is far less resource-intensive.
By recent measurement, a single, GenAI prompt uses 10 times the energy as an internet search. However, the AI tool you use and the complexity of your prompt matter, with complex prompts using 9 times more energy than a simple prompt. And, when you prompt GenAI repeatedly to give you better and different answers, your energy use grows substantially and adds up. That’s why projections show gigantic global energy use increases as the data centers required for AI tool scale up.
At Pitt we actively select cloud vendors for our AI systems like AWS, which has data centers with very high performance per watt, makes chips that use 40 to 60% less energy than standard chips to perform the same task, purchases 100% renewable electricity, and prioritizes responsible water use. Data centers require water primarily for cooling. When temperatures rise, evaporative systems pass air through water-soaked pads to regulate equipment temperatures. AWS addresses this by using recycled wastewater instead of drinking water at data centers worldwide and has committed to becoming “water positive” by 2030, meaning it will return more water to communities than it consumes. As with all Pitt Purchasing efforts, ensuring that sustainability is embedded in selecting AI tools and cloud partners helps ensure Pitt AI efforts align with local and global sustainability and climate goals.
Regardless, our hope is that if you use GenAI effectively, you have the potential to save far more power than you’ve used.
Creating AI-Based Climate Solutions
Beyond answering your questions and creating funny images, AI can be a tool to tackle our most pressing climate challenges. For example, here at Pitt, researchers are using AI to design more sustainable concrete. Given that cement production accounts for 7% of global carbon emissions, making greener concrete can help reduce global greenhouse gas emissions.
Pitt researchers are also advancing Responsible Data Science, including applying AI in a responsible manner to key climate, social, and health challenges. AI is being used to optimize power grid efficiency and create predictive models for environmental risk. There’s real potential for AI and generative AI to support systems-level solutions to climate challenges.
These conversations are happening across the University — at events like the Global Innovation Summit 2025 co-hosted by Pitt and CMU and through new academic offerings like Pitt’s Health Informatics and AI undergraduate degree program.
Carbon Neutral by 2037: Our Institutional Commitment
Having a sustainable AI strategy is an integral part of Pitt’s commitment to reaching carbon neutrality by 2037, as outlined in the Pitt Climate Action Plan (PittCAP). This goal shapes real decisions about how the University invests in technology and infrastructure, including AI.
The PittCAP and Pitt Sustainability Plan balance equity, environment, and economics, which helps advance sustainable solutions that work for people, the planet, and institutional and economic realities. At Pitt Digital, we actively consider the environmental impact of GenAI tools when making purchasing decisions and prioritizing energy-efficient equipment and vendor tools.
Be Thoughtful About Your GenAI Use
So, what can you do? Here are some tips to use AI tools efficiently:
Are you asking the right questions? Being more deliberate about your prompts and usage can reduce wasted computing resources. All unnecessary words influence GenAI runtime and energy requirements; for example, using “Please” or “Thank you” in your prompt require more processing power (and energy).
Reduce unnecessary communication. Can GenAI help summarize a lengthy email chain or quickly find a solution, saving the energy of extended back-and-forth communication?
Avoid waste. Are you using GenAI to write first drafts that you ultimately discard? If so, you might be literally burning energy on work that never sees the light of day.
Use the right tool for the task. Before using GenAI at all, first ask: “Is this the right tool?” For instance, it would be a poor choice to use AI for a simple calculation instead of a calculator. Even for specialized tasks, there might be smaller, purpose-built AI tools designed for specific fields that use significantly less energy than more general-purpose models like ChatGPT.
These questions put you in control of making AI use meaningful instead of just convenient. AI impacts us all; each of us needs to be mindful about our digital usage.
Moving Forward Together
Responsible innovation means being transparent about the potential pressure points while staying open to technological advancements that can bring us closer to a truly sustainable world. It's not always easy to strike a balance, but by starting these conversations, we are preparing ourselves to evolve as quickly as our technology does. At Pitt Digital, we are committed to building a future that is determined not just by what technology can do, but by the values and priorities we bring while using it.
Learn More
Want to dig deeper into responsible AI use? Check out these resources:
- Science News: “How much energy does your AI prompt use? It depends.”
- What Uses More? Tool to compare AI use to other common activities. Developed by the University of Maine.
- “Should I Feel Guilty Using AI?” video by Simon Clark
- Machine Ethics Podcast on “DeepDive: AI and the Environment”
- Environmental & Energy Study Institute: Data Centers & Water Consumption
—Pitt Digital
