THE ENERGY PARADOX

As Al reshapes the world, can it curb its own environmental impact?

- 

ON FEBRUARY 10-11, the world gathered at the Paris AI Action Summit to address the surging energy appetite of AI. The third instalment of the summit brought together governments, AI companies, civil society, and experts—highlighting concerns over Al’s escalating energy consumption as nations compete to lead the AI revolution.

The key outcome was the announcement of an Observatory on Energy, AI and Data Centres. Expected to launch in April 2025 under the International Energy Agency (IEA), the observatory will compile global data on Al’s electricity demand while tracking Al applications in the energy sector.

AI presents a contradiction. While hailed as a breakthrough capable of optimising energy grids and combating climate change, its own power consumption is a growing problem. “AI has the potential to reduce environmental harm by improving energy efficiency, optimising resource use, and integrating renewable energy sources,” says Maria Basso, head of AI applications and impact at the World Economic Forum. However, she cautions that large-scale AI models consume vast energy, adding to carbon emissions. AI applications like ChatGPT rely on generative pre-trained transformer (GPT) models—complex neural networks that mimic human cognition. These models require power throughout their lifecycle, spanning five phases: planning and data collection, model development, training, deployment, and ongoing maintenance.