The Pervasive Impact of ChatGPT - Shouldn't We Discuss Its Hidden Environmental Impact?
In the rapidly evolving world of technology, AI models like ChatGPT are making significant strides, revolutionizing various aspects of our lives. However, a lesser-known fact is the environmental impact these models have, particularly in terms of energy and water consumption.
Mark Zuckerberg of Meta has suggested that AI chatbots could potentially replace real-life friends, but the energy required to power these models is a cause for concern. According to the International Energy Agency, a single ChatGPT query consumes roughly 2.9 to 3 watt-hours (Wh) of energy, which is nearly 10 times more energy than a typical Google search.
This high consumption is due to the need for running complex AI models continuously in data centers, which require significant computational resources and electricity. AI inference, or actually running the model queries, accounts for over 80% of AI’s total energy consumption.
Moreover, the cooling systems used in data centers to manage the heat output from intensive computation use water more efficiently than traditional air cooling. As a result, AI data centers consume more water compared to conventional data centers, although exact per-query water usage figures are not readily available.
The emergence of AI models like ChatGPT is driving sharp increases in both electricity and water consumption in data centers, contributing to a growing environmental footprint compared to standard internet search queries. This underscores the importance of ongoing efforts to improve AI hardware efficiency, cooling technologies, and energy sourcing.
Young people, in particular, are increasingly concerned about sustainability and the environment, especially under the administration of former President Trump, who was criticized for undermining the country's climate ambitions. Awareness of the resource usage of AI models like ChatGPT is crucial for conservation efforts.
Shaolei Ren, an associate professor at the University of California, Riverside, who studies AI's impact on climate, emphasized that if people are not aware of the resource usage of AI, they cannot help conserve resources. The administration of President Trump has also been criticized for sapping the US power grid, which has historically been under-invested in and under-resourced.
As AI chatbots like ChatGPT are being used for various tasks, including drafting texts and therapy, it is essential to consider the environmental impact and strive for sustainable practices in AI development and usage. Companies like Microsoft and Meta are already investing in nuclear energy to expand their energy capacity for AI usage, but critics argue that this comes with its own set of potential problems for people and the environment.
The Washington Post, in collaboration with researchers at UC Riverside, conducted an analysis in 2024, finding that data centers are consuming large amounts of resources. As AI models continue to evolve and become more integrated into our lives, it is crucial to balance technological advancement with environmental responsibility.
Scientists in the field of environmental science are increasingly focused on the environmental impact of AI models like ChatGPT, as their energy consumption, particularly during AI inference, contributes significantly to both electricity and water usage. In light of this, it is crucial for developers to prioritize energy efficiency and sustainable practices in AI research and application, to mitigate the growing environmental footprint caused by these models.
