AI & Climate

The discourse surrounding AI often neglects to address its significant contribution to carbon dioxide (CO2) emissions, which rivals or even surpasses traditional industries like air travel. The proliferation of AI technologies, particularly data centers and computational infrastructure, has led to a substantial increase in energy consumption. These AI systems require immense computational power, often provided by energy-intensive data centers operating around the clock, thus increasing the carbon footprint of AI technologies.

For instance, Raghavendra Selvan's research has shed light on the energy cost of running a single prompt for ChatGPT. It was found that one prompt consumes an average of 0.19 kWh of energy, equivalent to charging your mobile phone up to 40 times. This underscores the significant energy consumption associated with AI operations.

This aspect of AI's environmental impact is often downplayed with the dominating argument AI as a solution to climate challenges without acknowledging its own contribution to carbon emissions. While AI is touted as a tool for improving efficiency and effectiveness in various sectors, its environmental impact still needs to be thoroughly examined and addressed.

Sources

The Techno-Optimists of Climate Change: Science Communication or Technowashing?

AI's stigende CO2-aftryk er en udfordring. Her er, hvad vi kan gøre ved det