Generative AI's increasing presence in daily tasks masks a growing environmental burden. Training these AI models demands substantial computational power, leading to significant electricity consumption and subsequent carbon emissions. The energy needed to both train and operate these models places a strain on power grids. Furthermore, cooling the hardware requires vast amounts of water, potentially stressing local water supplies. The manufacturing and transport of high-performance computing hardware also contribute to the indirect environmental impact.
Each interaction with AI, such as a simple query, consumes more energy than a typical web search. The rapid expansion of AI applications amplifies these effects, making it crucial to address the environmental consequences. Some experts suggest AI could help mitigate greenhouse gas emissions through process optimisation, but the intensive resources required by generative AI offset these benefits. Data centres powering AI contribute significantly to global emissions, with projections indicating a substantial increase in the coming years.
While only a fraction of companies currently measure the carbon footprint of their AI activities, there's growing recognition of the need to integrate sustainability considerations into AI deployment. Balancing accuracy with environmental impact is essential, alongside exploring cleaner energy sources for AI infrastructure.
