We all have been talking about how Artificial Intelligence has changed our lives. This meteoric rise of AI has redefined boundaries of what machine can accomplish. However behind the scenes of this technological renaissance lies a growing concern ,one that is not as visible but is just as significant - the enormous energy consumption of AI.
Fact check - International Energy Agency predicts that Electricity usage by data centres will increase to between 620 and 1050 TWh in 2026 which is equivalent to energy demands of Germany.
1) Primary Factors contributing to high energy consumption of AI Systems
➡ Training the Model - To train a Deep Learning Model, you require vast amount of data. Processing and analysing these datasets consumes significant amount of energy
➡ Data Centres - AI Models are deployed and trained in the big data centres. These data centres not only powers the servers but also require substantial energy for cooling to prevent overheating.
➡ Inference of AI Models - This energy consumption continues into the inference stage where they are used to make predictions or performs tasks.
2) What are environmental implications of growing energy demand of AI?
➡ Increased Carbon Emissions - The power consumed by the the AI infrastructure emits Greenhouse Gases into the atmosphere and leads to Global Warming and Climate Change.
➡ Electronic Waste - Due to rapid advancement of AI, there is frequent hardware upgrades and often electronic waste is not disposed efficiently.
➡ Water Usage for Cooling - Increasing Data Centres has increased the need of water which is leading to depletion of water resources.
➡ Strain on Energy Resources - Due to the huge chunk of energy required for AI Infrastructure , the strain on different sources of energy is also increasing leading to massive shortages.
3) Steps to be taken for a Greener AI
➡️ Powering the AI Infrastructure with renewable energy sources such as Solar and Hydroelectric power.
➡️ Different carbon offsetting projects such as reforestation should be encouraged.
➡️ Transparency should be observed by different developers and researchers in their disclosure of energy consumption.
➡️ Proper disposal and recycling of E-Waste.
4) Measures taken by leading Companies to reduce the Carbon Footprints
➡ Facebook Data Centre in Lulea, Sweden uses 100 percent of renewable energy
➡️ Amazon has invested $2 Billion in Climate Pledge under which it aims to reach net-zero carbon in his business by 2040.