AI's Carbon Problem

The use of Artificial intelligence (AI) has risen in recent years and has proven to be incredibly useful in a number of different industries. However, the harm it causes the environment should not be overlooked. If we are to aim for a sustainable future, the U.S. must regulate the usage of AI.

Artificial Intelligence requires a ton of energy, emitting Greenhouse Gases (GHG) into the atmosphere, one being carbon dioxide (CO2). The Training stage of an AI model is incredibly energy intensive, which means a ton of carbon is released. For example, ChatGPT-3 was found to emit 502 tons of CO2 from training alone. The Inference stage can also emit a substantial amount of carbon dioxide into the atmosphere with frequent use and millions of users worldwide. 

Large corporations, such as Microsoft and Google, own and operate data centers all across the globe. These companies have decided to increase their use of AI technology, which increases their energy consumption and emits CO2. Microsoft reported since 2020, their emissions had grown 29%. The increase is largely due to new data centers designed to focus on AI technology. Additionally, Google said their GHG emissions had risen by 48% since 2019 from their energy usage in data centers. The report also noted the difficulty of reducing carbon emissions with the implementation of AI technology.

Increased carbon emissions from AI systems and data centers disrupts the global climate. GHG emissions heat up the earth, which harms ecosystems, shifts weather patterns, and raises sea levels. It is for these reasons the U.S. must put regulations on AI and pursue sustainable technology.