Computer Work
iStock Photo Courtesy of Discipleship Ministries
Editor’s Note: Just as we were finalizing this article, new research emerged[1] suggesting significant improvements in AI’s energy efficiency—specifically, ChatGPT queries consume around 0.3 watt-hours, much lower than earlier estimates. This timely update highlights how rapidly AI technology is evolving, reminding us that our understanding of its environmental impact is still developing.
How many Google searches equal the energy of a single ChatGPT query? It’s a question worth pondering—not for trivia’s sake, but for what it reveals about the energy we consume daily and the decisions we face as stewards of God’s creation, striving to balance our technological advancements with sustainable and just practices. Before answering this question, consider this: AI’s energy use isn’t just a technical detail—it’s a window into the broader challenges of balancing innovation, responsibility, and sustainability. To grasp its ecological impact, we must explore how AI fits into the larger narrative of our energy consumption, its transformative potential, and the role it might play in addressing global crises. By looking at the bigger picture, we can weigh its costs against its capacity to drive meaningful change.
To navigate this intersection of technology and stewardship, this article brings together perspectives that might seem in tension: two AI advocates who train church leaders in AI use and a stewardship director focused on sustainable ministry practices. We aim not to take sides in this debate, but explore how Christians might thoughtfully engage with these emerging technologies amid conflicting headlines about their benefits and risks.
The Energy Cost of AI
Artificial intelligence tools, such as Large Language Models (LLMs) like ChatGPT, and image/video generation tools like Midjourney, Adobe, and Sora are resource intensive. Training these models requires vast computational power that consumes millions of kilowatt-hours. Even individual queries, though seemingly small, are energy-heavy compared to more familiar technologies. The environmental impact of AI goes beyond energy consumption; data centers often depend on water-intensive cooling systems, creating additional challenges in regions where water scarcity is already a concern. This draws attention to the significant environmental costs associated with AI systems, urging us to examine how these technologies impact the planet while weighing their immense potential for positive transformation. Additionally, much of the energy powering these data centers currently relies on fossil fuels, and the resulting carbon emissions directly contribute to climate change, exacerbating extreme weather events and biodiversity loss. However, many companies are actively exploring smaller AI models and renewable energy sources to reduce this impact.
Putting AI’s Energy Use in Context
While AI’s energy demands are significant, they’re not unprecedented. Steel production, for instance, consumes far more energy. Producing just one ton of steel[2] requires anywhere from 625 to 6,805 kWh (kilowatt-hours), equivalent to 200,000-2.3 million ChatGPT queries. Likewise, other daily activities, like running a central air conditioner for an hour, use the same energy as one thousand AI queries. It might sound simple to reduce this to comparisons like "But running my air conditioner for an hour has tangible benefits when compared to a thousand generated images of a cat doing kung-fu or actor Will Smith eating spaghetti," but the authors of this article find that to be an over-simplification of the realities and potential benefits of these tools.
To provide a broader context, here’s a comparison of energy use across various tools and industries:
Abbreviations:
GWh: gigawatt hours
Wh: watt hours
KWh: kilowatt hour
Energy Comparison 1
UM Insight Screenshot from Discipleship Ministries
Energy Comparison 2
UM Insight Screenshot from Discipleship Ministries
United Methodist Social Principles
The United Methodist Social Principles emphasize sustainability, justice, and stewardship of the natural world. These values directly relate to energy use in artificial intelligence systems. The Social Principles advocate for sustainable energy policies that reduce reliance on fossil fuels and promote renewable energy sources like solar and wind. Additionally, they call for energy efficiency and conservation practices, such as reducing carbon footprints and conducting energy audits. Just energy policies are prioritized to ensure fairness in the distribution of energy benefits and burdens while minimizing harm to creation. These values align closely with the need to manage AI’s growing energy demands responsibly, encouraging both individual and systemic changes. The church should call for AI companies to be transparent in reporting their energy use, ensuring accountability and fostering practices that prioritize sustainability. To be clear, we believe the church should call ALL technology companies to do the same, not just the ones currently in the spotlight (AI).
AI as a Tool for Stewardship
United Methodists are committed to being good stewards of God’s creation, so the energy-intensive nature of AI presents a challenge. However, this same technology holds the promise of contributing to solutions that align with our call to care for creation. By fostering sustainable energy practices, supporting environmental monitoring, and promoting individual and systemic changes, AI's potential to create positive environmental impacts offers hope for the future. It’s important to note that while AI tools like ChatGPT consume more energy per use compared to traditional search engines, their overall impact should be considered in context:
AI's potential for energy savings: AI technologies have the potential to optimize energy use in various sectors, potentially leading to significant energy savings that could offset their consumption [PowerVine Energy, “Understanding Energy Usage in Daily Life”]. Furthermore, AI and machine learning technologies are already in smart home features that can lead to lower energy consumption. As this technology improves, it may be able to schedule power-consuming activities when energy demand is lower.
Comparison to everyday activities: The energy used by AI tools is relatively small compared to common household activities. For example, using ChatGPT for one hundred queries consumes roughly the same amount of energy as running an air conditioner for thirty minutes [EcoWatch, “Electricity Usage: How Many kWh Does a House Use?”].
Ongoing efficiency improvements: As AI technology advances and the training continues, efforts are being made to improve energy efficiency. Additionally, the release of China’s AI model, DeepSeek, is likely to put pressure on other AI companies to optimize their models and reduce the need for such high energy consumption.
Impact on infrastructure: The growing demand for AI is driving significant increases in data center energy consumption. Projections indicate that data centers could consume up to eight percent of total power in the U.S. by 2030, up from three percent in 2022 [PowerVine Energy, “Understanding Energy Usage in Daily Life.”].
Environmental initiatives: Major tech companies are setting ambitious environmental goals. Google aims to achieve net-zero emissions by 2030, while Microsoft targets being carbon-negative by the same year [PowerVine Energy, “Understanding Energy Usage in Daily Life.”].
Beyond Energy: AI’s Broader Contributions
AI’s benefits extend beyond climate solutions. Advances in healthcare, for instance, highlight the potential of AI to improve lives. Recent developments include AI-powered tools for quick, no-contact blood pressure and diabetes screenings, which could revolutionize early diagnosis and accessibility to preventive care. Additionally, AI has demonstrated significant capabilities in fraud detection, helping federal agencies recover $1 billion in fraudulent claims in a single year and showcasing its potential to enhance accountability and resource management.
AI-powered diagnostics can identify diseases earlier, leading to better outcomes. Predictive models optimize hospital resources, while robotic tools assist in surgeries with precision impossible for human hands alone. By alleviating suffering and extending care, AI contributes to the common good in ways that resonate with Christian values of compassion and healing.
Is It Worth It?
So, back to our original question: how many Google searches equal the energy of a ChatGPT query? A single ChatGPT query consumes about ten times the energy of a Google search. Is it worth it?
I lean toward the optimistic view that the benefits outweigh the costs. One practical way to reduce these costs is by opting for smaller AI models (often called “mini” models). These models require less computational power and energy to operate, making them an efficient alternative for many applications without sacrificing functionality. By intentionally selecting tools that balance capability with energy efficiency, we can ensure that AI serves its purpose responsibly and sustainably. When employed thoughtfully—to advance knowledge, improve lives, and care for creation—AI becomes a powerful tool for stewardship and innovation.
When viewed through the lens of societal impact, it is worth reflecting on the energy use and benefits of media and tools that we prioritize, whether it be bitcoin and other cybercurrencies, social media, AI, and other major uses of energy. It is of note that AI technologies are already finding potential new drugs for antibiotic-resistant bacteria (something that for one of the authors of this article who has cystic fibrosis, could be life changing).
That same author uses AI Large Language Models to help build plans and direction for eliminating invasive plant species wreaking havoc to the ecological landscape of his farmland. He is using its expertise to steward more than a hundred replacement plantings of native plant and tree species to restore the ecology of his land.
Our call is not to reject AI but to wield it responsibly, ensuring it serves the greater good while minimizing harm.
As we navigate this new frontier, let’s remain wide-eyed stewards about the costs as well as inspired by the possibilities, trusting that with careful discernment, technology can be a force for good in God’s world.
Reflection Questions:
- How do we determine whether the benefits of AI outweigh its environmental costs? What principles from Christian stewardship can guide our discernment?
- Compared to other high-energy technologies like bitcoin and air travel, does AI’s potential for improving lives justify its energy use? Why or why not?
- In what ways can the church advocate for responsible AI development while also encouraging its use for social good?
- How does our perception of AI’s value change when we consider its potential in healthcare, environmental conservation, and social justice?
For Additional Reading
Agway Energy. “Average kWh per Day” [8/24/24], https://www.agwayenergy.com/blog/average-kwh-per-day/.
NRG Clean Power. “How Many kWh Per Day Is Normal?” [11/18/2024], https://nrgcleanpower.com/learning-center/how-many-kwh-per-day-is-normal/.
Green Logic. “The Top 5 Biggest Users of Electricity in Your Home.” [1/3/2024]. https://greenlogic.com/blog/the-top-5-biggest-users-of-electricity-in-your-home.
EcoWatch. “Electricity Usage: How Many kWh Does a House Use?” [12/18/2024]. https://www.ecowatch.com/electricity/how-many-kwh-does-a-house-use.
PowerVine Energy. “Understanding Energy Usage in Daily Life.” [10/1/24], https://www.powervineenergy.com/blog/understanding-energy-usage-in-daily-life.
U.S. Energy Information Administration. “Today in Energy: Data Center Energy Use.” [2/21/20]. https://www.eia.gov/todayinenergy/detail.php?id=42915.
American Heart Association. “AI-Powered Tool May Offer Quick, No-Contact Blood Pressure and Diabetes Screening.” [11/11/24]. https://newsroom.heart.org/news/ai-powered-tool-may-offer-quick-no-contact-blood-pressure-and-diabetes-screening.
CNN. “AI Helped the Feds Catch $1 Billion of Fraud in One Year.” [10/17/24]. https://www.cnn.com/2024/10/17/business/ai-fraud-treasury/index.html.
[1] https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use.
[2] http://large.stanford.edu/courses/2016/ph240/martelaro1/?t and https://rmi.org/wp-content/uploads/2019/09/green-steel-insight-brief.pdf?t.
[3] https://www.statista.com/statistics/1536926/ai-models-energy-consumption-per-request/ and https://www.heise.de/en/news/ChatGPT-s-power-consumption-ten-times-more-than-Google-s-9852327.html?t.
[4] https://www.statista.com/statistics/1536926/ai-models-energy-consumption-per-request/ and https://www.heise.de/en/news/ChatGPT-s-power-consumption-ten-times-more-than-Google-s-9852327.html?t.
[5] https://greenly.earth/en-gb/leaf-media/data-stories/the-hidden-environmental-cost-of-social-media?t and https://greenspector.com/en/social-media-2021/?t.
[7] https://bkvenergy.com/blog/how-much-electricity-does-refrigerator-use/?t.
[8] https://naturesgenerator.com/blogs/news/how-many-watts-does-a-light-bulb-use.
[9] https://www.ck12.org/section/planes/.
[10] https://blog.ecoflow.com/us/how-much-kwhs-does-average-american-household-consume/?t.
Scott Hughes is the Executive Director of Congregational Vitality & Intentional Discipleship, Elder in the North Georgia Conference, M.Div. Asbury Theological Seminary, D. Min. Southern Methodist University, co-host of the Small Groups in the Wesleyan Way podcast, creator of the Courageous Conversations project, and facilitator of the How to Start Small Groups teaching series.
This article is republished with permission from Discipleship Ministries' website.