The European Union is in the midst of crafting the world’s first comprehensive law for Artificial Intelligence (AI). This new law wants to make rules for AI. It is like a guidebook to control AI and make sure it does not cause problems for people, society or the environment. While discussions revolve around these risks, there is a significant environmental concern that often goes unnoticed: the potential increase in energy consumption due to the widespread use of AI in our daily lives.
AI helps fight climate change by making energy grids work better, predicting changes in the weather and keeping an eye on agreements about the climate. However, the infrastructure required to run AI is energy- and resource-intensive. For instance, training large language models like OpenAI’s GPT-3 demands substantial electricity for data centers, further requiring significant water usage for cooling.
AI does not just affect the environment by using energy. It is now a big part of our digital lives, helping businesses make things better and changing how we search, shop as well as do daily stuff. This means we use more energy and we are trying to use less right now.
AI can change how we act and use energy, even though it does not use energy directly. AI-powered systems, like smart home technologies, promise energy efficiency gains but may lead to a “rebound effect” if increased comfort expectations and additional device purchases offset the intended energy savings.
Efforts to account for and reduce the direct carbon footprint of AI infrastructure are important but insufficient. AI developers need to think about how we act when they design it. This helps avoid problems and keeps us from using too much energy in a world that uses AI a lot.
