ChatGPT has quickly become a fixture in many of our daily lives — whether that’s asking the generative AI chatbot for party-planning assistance, dating advice, therapy or Studio Ghibli-inspired art. Over 400 million people across the globe use ChatGPT each week, with 122.6 million daily users.
While asking ChatGPT for help can feel harmless (and a smart way to save time and focus on the things that we love), using the AI chatbot can have a colossal impact on the environment. According to Professor Gina Neff, Deputy CEO of Responsible AI UK, ChatGPT uses an immense amount of energy; the data centres used to power it consume a higher amount of electricity per year than 117 countries.
Amidst global efforts to reduce carbon emissions, the surge in AI usage forces us to confront important questions: What impact are these technologies having on the planet and how are our actions contributing?
ChatGPT’s environmental impact
A ChatGPT search uses an estimated five times more electricity than a web search, but the generative AI has other often overlooked environmental impacts. We break these down below.
Training
Training ChatGPT requires a huge amount of electricity, resulting in high carbon dioxide emissions. In 2021, scientists from Google and the University of California estimated that training a model like ChatGPT consumed 1,287 megawatt hours of electricity — generating around 552 tons of carbon dioxide.
All machine-learning models are trained, but what is different about generative AI models is that their energy use fluctuates throughout the training process. This means that power grid operators must absorb the fluctuations in order to safeguard the grid, often via diesel-based generators (which are known to produce toxins such as carbon dioxide and nitrogen oxide).
What’s more, the training process is continual, as ChatGPT is constantly in development, meaning that more and more energy is required. “Generative AI models have an especially short shelf-life, driven by rising demand for new AI applications,” Noman Bashir, Computing and Climate Impact Fellow at MIT Climate told MIT Technology Review.
Water
An often overlooked environmental impact of AI is the amount of water needed to cool the hardware used for training. Data centres take water from the local water supply to cool down the excess heat generated.
ChatGPT uses about 39.16 million gallons of water daily, the equivalent of everyone in Taiwan flushing the toilet, according to Business Energy UK.
“When we think about the environmental impact of generative AI, it is not just the electricity you consume when you plug the computer in. There are much broader consequences that go out to a system level and persist based on actions that we take,” Elsa A. Olivetti, lead of the Decarbonization Mission of Massachusetts Institute of Technology’s Climate Project told MIT News.
Queries
Once the training process of a generative AI model is complete, the energy demands continue. Research indicates that a ChatGPT query consumes approximately five times more electricity than a web search.
“The demand for new data centers cannot be met in a sustainable way. The pace at which companies are building new data centers means the bulk of the electricity to power them must come from fossil fuel-based power plants,” Bashir told MIT News.
However, people do not tend to consider the environmental impact of their actions when using the chatbot for menial tasks, like recipe ideas or book summaries.
Bashir adds: “The ease-of-use of generative AI interfaces and the lack of information about the environmental impacts of my actions means that, as a user, I don’t have much incentive to cut back on my use of generative AI.”
