Razaq, now a freshman at UIndy, was a Reporter for YEPT in 2023/2024 when she also wrote for Elkhart High School’s ‘The Pennant.’

In schools and universities, conversations about artificial intelligence (AI) often focus on its morality — questions about academic integrity, and ethical dilemmas dominate the discussion. But one very real issue gets left out of these discussions: the environmental impact of AI.

While sites like ChatGPT might seem harmless, they consume large amounts of energy and leave a significant carbon footprint. In a time where climate change is one of humanity’s most pressing challenges, ignoring the environmental cost of AI is a mistake.

AI systems like ChatGPT depend on machine learning models that require massive computational power. Training these models takes enormous amounts of energy. For example, training GPT-3, the foundation of ChatGPT, produced an estimated 284,019 kilograms of carbon dioxide (CO2). To put this number into perspective, it is about the same as the lifetime emissions of five cars. Most of this energy comes from data centers, which are still heavily reliant on fossil fuels. These centers alone account for about 1% of the world’s electricity use and 0.3% of global CO2 emissions, and it is a number that’s only rising every year.

And the energy demand doesn’t stop once the models are trained. Keeping AI systems running and accessible for millions of users worldwide requires constant power. Data centers hosting AI applications are estimated to increase their energy use by 4% annually as cloud-based AI becomes more popular. In regions where electricity grids still depend on coal or natural gas, this energy consumption directly translates to more emissions, only further fueling the climate crisis.

It’s not just the electricity that’s a problem. The hardware needed to train and run AI models — like GPUs and servers — depends on rare earth metals such as lithium, cobalt, and nickel. Extracting these materials often involves destructive mining practices that harm ecosystems and pollute water supplies. On top of that, the constant need for newer, faster hardware creates mountains of electronic waste, much of which ends up in countries with weak environmental protections such as Nigeria, India, and Vietnam.

One area where the expansion of data centers is actively supported is Indiana, particularly from the state’s governor, who has emphasized their importance for economic growth and job creation. Governor Eric Holcomb stated, “Indiana is committed to being a central hub in the global economy of the future, and this latest announcement ensures Hoosier communities and talent will be key to widespread advancements in cloud and artificial intelligence technology.”

This support is reflected in the 2024 announcement of a new $1 billion Microsoft data center in Northwest Indiana, with additional plans potentially influencing 2025 bills and legislation that could streamline data center expansion. While such growth can boost local economies, it also raises concerns about the environmental impact, especially the increased energy consumption and the demand for rare earth resources required to maintain these facilities. 

Despite all this, the environmental cost of AI rarely gets the attention it deserves. The tech industry loves to highlight the benefits of AI but it often stays silent about the environmental toll. For instance, OpenAI, the company behind ChatGPT, hasn’t disclosed the full impact of its operations, which shows how little transparency exists in the field. Without clear data, it’s hard for governments or consumers to hold companies accountable or push for more sustainable practices.

There are solutions to this problem, but they require significant effort. Transitioning data centers to renewable energy is a crucial step. Google, for example, has pledged to make its data centers carbon-free by 2030. More companies need to follow their lead. Researchers also need to develop energy-efficient AI models that can reduce the computational power required without sacrificing performance. Techniques like model compression and optimization offer promising paths forward.

Education also has a role to play. Universities and schools need to expand the conversation around AI to include sustainability alongside ethics. Students should understand that technological progress comes with a responsibility to minimize harm, both to people and the planet. If future AI developers grow up with this mindset, we’ll be better equipped to balance innovation with environmental responsibility.

AI has incredible potential to improve lives and transform industries, but we can’t afford to ignore its environmental cost. Programs like ChatGPT are energy-hungry and resource-intensive, and are contributing to a growing carbon footprint that we can no longer overlook. Addressing this issue will take collaboration from tech companies, policymakers, and educators alike. But by shining a light on the environmental impact of AI, we can push the industry toward greener practices and ensure that innovation doesn’t come at the expense of the planet’s future.