Three clicks to write a play or create a website: artificial intelligence systems (AI) may seem like magic, but their functions aren't environmentally neutral. They’re intensive consumers of water and electricity.
AIs don't drink water, but data centres, where artificial intelligence systems are trained, use a lot of it to cool their servers. This is only one piece of the puzzle when it comes to digital water consumption.
AIs such as ChatGPT and Bard consume far more water and energy than your typical internet search. According to a preprint from the University of California, Riverside, a conversation with ChatGPT consumes around 50 cl of water, the equivalent of a small plastic bottle. With nearly 1.5 billion users a month, that adds up fast.
Fluctuating CO2 emissions
Data centres, essential for AI training, account for almost 1% of the world's energy consumption. This figure is set to rise over the next few years.
But these centres emit varying amounts of CO2 depending on whether their base country produces its electricity from coal or gas, for example, or from renewable energies. In a paper funded by Microsoft and the Allen Institute for Artificial Intelligence, researchers showed that by modifying the training location of an AI, it was possible to reduce the operation's CO2 emissions by 75%.
There's also a lot of talk among digital companies about "following the sun", i.e. changing AI training locations throughout the day so as to be able to use solar energy continuously. Optimising AI training locations could be an important means of limiting their impact on the environment.
More concretely, an AI like Bloom, a completely open-source equivalent of ChatGPT developed as part of the BigScience research project, generated the equivalent of 25 tonnes of CO2 during its training. This was despite the fact that most of the energy used came from nuclear power, and was therefore carbon-free. For GPT-3, which runs ChatGPT, its estimated carbon footprint is 20 times higher, equivalent to around 300 return trips from Paris to New York by plane.
AIs supporting the planet
Despite their environmental impact, AIs have a role to play in the fight against global warming. For example, they can help meteorologists predict extreme weather events, or optimise industrial processes to reduce CO2 emissions.
The particular challenge for developers today is transparency. Most publicly available AI models do not reveal where the models were trained, or the carbon cost of using them.
When users do have this information, they can make their own informed choices. For example, before bombarding ChatGPT with basic questions, we can just do a simple browser search, which is less power-hungry.
To find out more, watch our video above.