By Denise Calderon, Oct. 29, 2024
Every day artificial intelligence becomes a bigger threat as more energy is consumed and the carbon footprint worsens. Regardless of its extraordinary features, the environmental impact hurts our planet.
As AI continues to grow, the need for natural resources increases and poses harm to the environment. To create the current AI models, a significant amount of energy, water and raw materials are needed. Just because it is easily accessible, and thus taken for granted, we have to be aware of the long-term implications that AI will continue to have in the future.
One of the main concerns is AI’s energy consumption. According to an article published by Yale Environment 360 there are about 9,000 to 11,000 cloud data centers in the world. All of these centers require a lot of energy to run and create AI features. The International Energy Agency reported that by 2026, data centers around the world will double their energy consumption compared to the use in 2022. This will make the consumption increase to approximately 1,000 terawatts which is almost equal to Japan’s entire energy consumption.
This overconsumption of energy is harming the planet because much of it is coming from fossil fuels, which emit greenhouse gasses that are responsible for global warming. The IEA reported that one ChatGPT request consumes 10 times the electricity than a Google search.
CPP geology professor Nicholas Van Buer said AI has become “a big waste of energy,” and he said it could become an even bigger waste in the future.
“Invest in people, not machines,” said Van Buer. “Right now, AI is only about 2% of global energy use. So, in some ways, it’s not a huge problem yet, but it could be headed in that direction.”
Although Van Buer said AI’s global energy consumption is currently minimal, he is weary of what AI models could do to the planet as previous studies and reports have predicted.
Another environmental concern associated with AI is the amount of water being used to run all the machines and infrastructures needed to produce it. According to Yale Environment 360, about half a liter of fresh water is used in one session of questions and answers forGPT-3 that roughly contains about 10 to 50 responses.
The total amount of water used for computing servers and energy generation for thermoelectric power plants at Google, Microsoft and Meta totaled 2.2 billion cubic meters (or 581,2 billion gallons) in 2022. This is equal to the total annual amount of water withdrawal in Denmark, a country with a population almost 6 million people.
Our water supply is limited and unevenly distributed across the world. Freshwater scarcity is one of the biggest problems that everyone is facing because of the rapid population growth and droughts. This has already affected about 4 billion people across the globe. If there is no change, it is predicted to be a bigger issue by 2030.
“For each community, it all depends on what other use the water would be put to if it wasn’t going to AI,” said Van Buer. “There are definitely some communities where the water could be put to other, better uses than AI.”
He also emphasized how this water could be used for homes.
“A single big-data center might use 1,000 ‘acre-feet’ of water per year, which is about 300 million gallons, ” said Van Buer. “That water could be used to serve a lot of homes.”
AI should not be the most important concern for these multimillion-dollar corporations, but instead, they should be investing in people. Water, and many natural resources. should be benefiting humans, not computers trying to imitate human behavior.
The electronics used at the cloud data centers require large amounts of scarce materials from the Earth; to make 2 kg computer they need about 800 kg of raw materials (about 4.4 pounds). Also, the microchips used require rare earth elements. This can be problematic because most of the time, these materials are mined in destructive ways. These data centers also produce large amounts of electronic waste that can contain hazardous substances such as mercury and lead.
Van Buer said AI could possibly become a tool to help solve the crisis of raw material shortages like minerals.
“There is a huge shortage of people trained in geosciences, and a bunch of AI startups are claiming that they will be able to step in and fill this gap with AI,” said Van Buer. “Based on the geology and AI literature I have been a scientific peer reviewer for, I don’t think AI is anywhere close to being as competent as a real geologist yet. It’s just that computer scientists can’t tell the difference.”
Despite the advancements of AI and the potential it has to replicate human jobs, it is still not developed enough to fully rely on.
AI is a fairly new invention, and there is still much more to discover; but the environmental harm that it is creating can already be seen in the few years that it has taken off.
The future of AI could be environmentally friendly to the planet and even offer solutions for our current problems. For instance, implementing smart home systems can reduce gas consumption by around 5% by controlling heating and appliances according to The Conversation. Although the research behind these solutions has not been fully tested, these are steps in the right direction for the future of AI.
AI is inevitable and is here to stay, but it is up to professionals to keep this technology running in a way that will actually benefit our planet rather than themselves. These kinds of solutions are possible for the future if big corporations would focus more on human needs rather than the need to make billions of dollars at the expense of our planet.
Feature graphic by Lauren Wong
