Is AI destroying our planet?

AI Photo by Beyzaa Yurtkuran

It’s the new piece of technology that has people excited and scared in equal measure. Many are saying AI is our future and it’s best to embrace it rather than fight it. With it now being used, not just in business but also in everyday life - emails, letters even wedding speeches! You can’t Google anything anymore without AI providing an overview. So gone are the days of scrolling and finding information yourself - and using books for research is a distant memory.

But what many people don’t know is the environmental impact AI is having. The development and deployment of this new software requires massive energy consumption and water usage which contributes to increased carbon emissions and strains water resources. So, let’s take a deeper look into how AI is affecting our environment:

Water Consumption

According to a leading scholar on AI, the AI water usage already exceeds the water usage of the biggest beverage companies in the world! This is due to the vast amounts of water required by the data centres to cool down the servers due to the energy intensive computing process which generates a considerable amount of heat. If we continue on this trajectory, it is estimated that global AI-related infrastructure may soon consume six times more water than Denmark (with a population of 6 million) in 2027.

Carbon Emissions

Every time we ask AI to work for us, large data centres make it possible. The massive energy consumption that is required to train and run AI models often relies on fossil fuels for electricity which leads to greenhouse gas emissions. A study by Google and the University of California reported that training a GPT-3 (Generative Pre-trained Transformer 3, used to generate human-like text) resulted in 552 metric tons of carbon emissions - the equivalent of flying from Australia to the UK more than 30 times!

Energy Consumption

Data centres which house the servers for AI are estimated to consume 1-2% of the world’s electricity which is predicted to rise. Training AI models such as the GPT-3 requires massive amounts of electricity and computing power. It is estimated that large models like the GPT-3 use 1,300 megawatt-hours (MWh) of electricity which is the annual consumption of 130 homes. A simple request made through ChatGPT (a common text-generating tool), whether writing an essay, an email or a letter, consumes 10 times the electricity of a Google Search. The expeditious growth of AI is increasing the demand for electricity and therefore putting stress on energy grids. 

The Future of AI

AI is continuing to expand and develop at an alarming rate, aiming to have an impact on everything from healthcare, manufacturing to everyday life. But the consequences of this advanced technology could outweigh the benefits. It is worth noting that it does hold the potential to revolutionise sustainability by offering solutions to climate change, resource management and biodiversity conservation. But, at present, it could cause a lot more damage than good.