Recently, there’s been much discussion about the potential benefits that artificial intelligence can bring to climate change regulation.
For example, advanced technology, such as satellite data, is being used to identify large emission events — (see ELM’s recent methane rule finalization coverage here and ELM’s previous AI coverage here). AI also is being used to monitor rising sea levels along the United States’ coastlines — (see ELM’s previous sea level coverage here).
Less consideration, however, has been given to the potential adverse impacts these new technologies can have on the environment, including their potential drain on our natural resources.
AI has been touted as a potential key to solving the world’s water problems. For example, AI-powered biosensors have the ability to detect toxic chemicals in drinking water, and smart irrigation can optimize water usage on farmland. Ironically, AI consumes a significant amount of water in order to operate. Notably, in 2022, Google’s data centers used over 21 billion liters of potable water in 2022, a 20-percent increase from 2021. Similarly, Microsoft’s use of both water and electricity increased by one-third in 2022.
Unfortunately, this upward trend is unlikely to decrease anytime soon. While one Google search requires approximately half a milliliter of water in energy, ChatGPT, an AI-powered natural language processing tool, uses approximately 500 milliliters of water for every 5 to 50 prompts. In addition, it’s estimated that technology firms will use an estimated 4.2 to 6.6 billion cubic meters of water to run and cool data centers by 2027.
Moreover, AI requires the use of lithium, among other carbon-intensive materials. Historically, water, particularly groundwater, has been a vital component of lithium mining, as it is used for processing metals and suppressing dust — (see ELM’s previous groundwater coverage here).
AI’s use of water isn’t the only major environmental concern. AI technology also produces a significant amount of electronic waste (“e-waste”), such as servers, semiconductors, and microchips. It is estimated that approximately 70 percent of the toxic waste in landfills is attributable to e-waste. Additionally, fossil fuels are also frequently used to provide energy to run AI systems, which can lead to increases in greenhouse gas emissions.
As a result, more oversight and regulation of AI’s impact on the environment is likely on the horizon. Therefore, before AI can even begin to work on finding solutions to the world’s environmental problems, perhaps it should start by solving its own.