Leveraging AI for Sustainability—and its High Environmental Costs

Beyond business operations, companies are now leveraging AI to help them reach their ESG goals. Next year, 500+ cities will use AI-powered digital twin technology to help manage their infrastructures. 

AI is also playing a major role in reducing the carbon footprint of the real estate industry, which accounts for 60 percent of major cities’ emissions. AI can empower eco-friendly building retrofits by analyzing construction and maintenance documents at an otherwise unattainable scale. AI is also optimizing the performance of the energy grid, among other environmental benefits.

AI solutions come with an environmental cost that businesses must consider. The underlying technology supporting these AI solutions consumes large amounts of energy. For example, training the most recent ChatGPT model (GPT4) used the same amount of energy as 5,000 American households consume annually. 

A single request on OpenAI’s ChatGPT uses 9.5 times more energy than the average Google Search. Although generating text is not as energy intensive, creating one AI-generated image consumes the same amount of energy as charging your smartphone. As AI becomes increasingly incorporated into our workflows, it’s expected that by 2026, data centers will have the same energy footprint as Japan – the fifth largest energy consumer by country. 

What Businesses Should Consider for AI Innovation

To stay competitive, businesses need to adopt AI and can do so in ways that don’t compromise their ESG goals. 

A first step is to quantify the carbon emissions of their AI applications. Hugging Face, a leading platform for AI development, has introduced methods to measure the environmental impact of AI models. Additionally, consumers can now adopt models that have been trained on renewable energy or those that have lower environmental costs. 

Organizations must recognize that AI and sustainability, on the surface, have a tension that needs to be addressed.

About the Author

Angela Virtu is a professor in IT and Analytics at American University’s Kogod School of Business, where she teaches courses on technology, data analytics, machine learning, and artificial intelligence. As the AI Instructional Faculty Fellow, she leads Kogod’s AI strategy and spearheads faculty and staff training initiatives. Prior to academia, Angela developed AI and machine learning solutions that bridged theoretical innovation with practical, ethical applications, driving organizational integration and measurable business impact.