“I’ve stopped using AI; it’s bad for the environment.”

This common sentiment seems to be the latest buzz surrounding artificial intelligence in the Upper School and around the world. Recent research cautioning excessive AI use has outlined the environmental impact of AI on two primary fronts: electricity generation and water consumption. But are these reports telling the whole story?

Here’s the rundown: the creation and operation of AI models are the undertakings of large data centers capable of enormous computational power. However, these data centers require significant energy to operate, often using electricity generated from fossil fuels, which emit greenhouse gases and contribute to global warming. Additionally, these centers require substantial water consumption to cool hardware.

So, AI and the environment are foes. Case closed, right? Incorrect. Many have jumped to quick conclusions based on the narrowly focused information they’ve received through their news feeds, but approaching this issue necessitates a greater level of nuance and recognition of the fact that so much remains unknown.

First, there is substantial disagreement on the energy and water consumption of AI, depending on the methods used for calculating current usage and predicted future usage as more data centers are built.

Many people make comparisons to put usage in context. One report claims that 300 ChatGPT queries use one gallon of water, while one hamburger requires 660 gallons of water. Additionally, American energy usage from Fortnite is greater than that of ChatGPT, and asking ChatGPT 180,000 fewer questions saves about a tenth of the amount of carbon dioxide emissions as switching to a hybrid car or a plant-based diet. Still, others suggest that as AI usage increases, the energy impact will be more dramatic.

However, increasing AI usage will also correspond with increasing renewable energy generation. On Oct. 7, a report published by Ember Energy found that, for the first time ever, renewable energies are poised to overtake coal power in the total share of global energy production in 2025. Similarly, the International Energy Agency recently reported that global renewable power capacity is expected to double between now and 2030.

What does this mean for AI? Well, it means that as the world gradually converts to more efficient, renewable energy sources, the electricity consumption of AI data centers will be gradually less harmful to climate change.

Finally, artificial intelligence could actually prove extremely beneficial in approaching environmental issues. In fact, its benefits likely already outweigh its harms. A research project from North Carolina State University compiled various uses and case studies of situations where AI is particularly beneficial for the environment. The infographic below shows what they found.

In the end, it’s actually reassuring that people are being cautious about their AI usage. However, students should acknowledge that there is no consensus agreement on whether or not AI has or will have a net positive impact on the environment. It’s essential to consider the complexity of the situation and avoid making bold conclusions.