Discover How AI is Transforming Our Environment: Key Insights and Impacts

Artificial intelligence (AI) uses a lot of energy and, unfortunately, it also increases pollution. Since 2020, major tech companies have contributed to 0.6% of global greenhouse gas emissions through AI. These emissions are mainly from data centers that run AI models.

For instance, just one question to ChatGPT uses about ten times more energy than a standard Google search. Moreover, cooling the computers that handle AI tasks requires a significant amount of fresh water.

However, there’s a flip side. Some studies suggest that certain AI applications might be less damaging to the environment than human labor. As technology advances, AI could also help in finding solutions to reduce its own environmental footprint.

Recently, during the wildfires in Los Angeles in January 2025, many people shared concerns on social media. Posts on X (formerly Twitter) claimed that AI is “speeding up the climate crisis” and depleting water resources. These discussions have been ongoing since AI technology became widely accessible, especially after ChatGPT launched in late 2022. A blog post from December 2022 discussed AI’s carbon footprint and raised awareness about its impact on the environment.

Training powerful AI models consumes a significant amount of electricity, leading to harmful greenhouse gas emissions. The actual water usage for cooling servers involved in AI still needs more research, but it’s clear that AI’s environmental impact is a topic of concern.

Different AI models consume varying amounts of energy. For example, research estimates that GPT-4, an advanced model, produces between a quarter to half a pound of carbon emissions to write a simple email. Data centers, which house the servers for AI, are responsible for about 0.9% of energy-related greenhouse gas emissions globally.

Reports from 2024 show that companies like Google and Microsoft have seen sharp rises in their greenhouse gas emissions, partly due to AI investments. Google’s emissions increased by 48% from 2019 to 2023, while Microsoft reported a rise of 29.1% from 2020 to 2023, attributing these increases to their focus on AI technologies.

Scientists are exploring the long-term climate impact of AI, focusing on metrics other than just carbon emissions, like public health implications from pollution. Some recent studies indicate significant public health costs related to data center operations in the U.S., expected to exceed $20 billion annually by 2030.

A common claim highlights that a ChatGPT query uses about ten times the energy of a Google search, which aligns with estimates from the International Energy Agency. Indeed, a typical Google search requires around 0.3 watt-hours, while ChatGPT uses about 2.9 watt-hours.

Regarding water usage, data centers do need water for cooling, but claims surrounding their effect on global freshwater resources are often exaggerated. A 2023 study estimates that GPT-3 requires approximately half a liter of water for every 10-50 responses. These figures may rise with newer models like GPT-4, which could use up to 3 liters of water for a single email.

Despite these environmental concerns, some supporters of AI argue it can lead to reduced human environmental impact. Some studies show that AI-generated writing and illustrating may produce less carbon than human efforts. However, the quality of AI output and its efficiency in terms of energy consumption are still under review and could change the overall picture.

AI could also play a role in combating climate change. The United Nations Environment Programme mentions using AI to monitor and manage emissions. However, AI’s potential for misuse in harmful industries, such as warfare, raises additional environmental concerns. AI-driven warfare can rapidly escalate conflicts, leading to widespread ecological damage.

In summary, the debate continues: Will AI ultimately harm or help the environment? It’s still unclear, and much depends on how AI is used and regulated moving forward.

Source link

greenhouse gas emissions, data centers, artificial intelligence, Google, Shaolei Ren, fossil fuel emissions, Google search, climate change, International Energy Agency, environmental impact, global energy consumption