The digital age and its advancements have brought about a multitude of benefits and conveniences, but it has also resulted in an increased environmental cost. This is particularly evident when examining the carbon footprint of online activities like Google searches and AI model usage. Unfortunately, many companies lack transparency about these environmental costs, making it challenging for consumers to make environmentally conscious decisions.

The carbon footprint of the internet is determined by two primary factors. The first is the resources that companies use to develop their tools. The second is the choices that consumers make in using these tools. It’s a cycle that feeds into the larger issue of environmental sustainability.

Among the tools contributing to the internet’s carbon footprint, AI models stand out. They consume significant energy, particularly during their training phase. Some companies, like Meta, use the equivalent of a city’s power for AI development. This high energy consumption for computing power, data storage, and hardware production is resulting in AI’s growing carbon footprint. This is a troubling irony, considering AI’s potential to help combat climate change.

However, strides are being made to mitigate this issue. Code Carbon, for instance, is assisting developers in estimating and reducing their tools’ carbon emissions. They’ve found that smaller AI models are more energy-efficient than larger ones. This discovery is a step in the right direction, demonstrating that modifications in AI model size can indeed reduce their environmental impact.

The largest AI training processes have been consuming vast amounts of computing power, contributing over 1% of global energy use and 0.6% of global carbon emissions. These worrying statistics highlight the urgency of implementing solutions to reduce the AI carbon footprint.

Solutions can be implemented at multiple levels. At the corporate level, companies can increase their transparency about the environmental costs of their products and services. On a larger scale, the introduction of governmental policies can regulate and limit the environmental impact of technology development. Finally, at the individual level, consumers can make more informed choices about their online activities, thus reducing their personal carbon footprints.

AI can improve our understanding of climate change and even reduce its own carbon footprint. However, making this potential a reality requires proper governmental policies and a commitment to transparency about the environmental costs of AI. As we continue to navigate the digital age, it is critical that we balance technological advancements with our responsibility to the environment.

Science4Data is committed to cut through greenwashing and measure real impact. Join the journey to a sustainable future. Your actions matter.