Today, AI is everywhere: Apple Intelligence is all the rage on iPhones and iPads, Google introduced AI Overviews for every search query, and there’s even Amazon Q, the AI assistant that helps you shop. While artificial intelligence had its humble beginnings about 70 years ago, completing tasks like playing checkers on its own, it now encompasses a wide range of technologies that analyze huge sets of data, detect patterns and trends, and even predict future outcomes.
While the benefits of AI to society are promising, a brand-new concern is coming to light: AI has a large carbon footprint, requiring a lot of energy and water. By 2026, the International Energy Agency predicts data centers—which power AI and cryptocurrency—could consume 4 percent of energy usage worldwide, the same amount of electricity required to power Japan. While there are ways to offset AI’s environmental toll, it will require widespread awareness and ongoing sustainability initiatives in order for us to use this advanced technology responsibly.
AI’s complex algorithms do an amazing job of analyzing data and performing tasks without requiring humans to program every step. However, this necessitates large data centers that use a significant amount of energy, which in turn produces heat that needs to be cooled down using a lot of water. According to a deep dive by Piktochart using the Machine Learning Emissions Calculator, every query or message you send to ChatGPT produces approximately 4.32 grams of CO2. Sending 16 queries uses the same amount of CO2 as boiling an electric kettle, while 139 queries is equivalent to one washing machine cycle. A researcher at the University of California, Shaolei Ren, calculated that as few as five ChatGPT searches use enough water to fill a 16-ounce bottle.
In comparison, a regular Google search uses just 0.2 grams of CO2. Now, think about how many people use ChatGPT every single day. OpenAI reached 300 million active weekly users last December, with a billion messages sent every single day. If just one message costs 4.32 grams of CO2, that means ChatGPT alone uses well over 4 billion grams daily.
Apple's Unchanging Carbon Neutrality Claim
Apple is one of the five major AI-related emitters alongside Google, Microsoft, Meta, and Amazon. Despite being a leader in AI, Apple still claims it is on track to be carbon neutral by 2030, a goal announced in July of 2020. The thing to keep in mind about carbon neutrality, though, is it doesn’t mean a company produces no carbon. A carbon neutral company tries to lower their emissions while also working on carbon offset—meaning they absorb just as much carbon as they emit. To reach its goal, Apple uses renewable energy whenever they can, plus works toward restoring natural environments that absorb carbon, such as forests, grasslands, and wetlands.
Apple also uses AI technology to make their production greener. At the 2024 China Development Forum, Tim Cook explained how Apple uses AI in the recycling process. The company’s AI-powered recycling robot Daisy takes apart returned iPhones and either properly discards them or reuses them in new devices. Apple claims just one recycling robot unit can disassemble 1.2 million devices annually. With Daisy’s help, Apple refurbished almost 8 million Apple devices in 2018 alone. The most impressive material that can be recycled is the battery, which contains the finite rare earth mineral cobalt.
Can Artificial Intelligence Fix the Environmental Crisis?
While the average person may have used AI to generate a funny image or write a text, it can be used on a much larger scale. AI is showing promise in helping humans better monitor environmental problems, optimize our agricultural practices, create a model for climate change, reduce emissions from transportation, and use energy more efficiently. So, even though powering AI uses a massive number of resources, it can potentially solve some of the most pressing environmental concerns we face today.
“As with any global environmental concern, everyone can take small steps to decrease their AI carbon footprint.”
It’s also being applied in academia. For example, in early 2024, AI was able to decipher some ancient Roman characters in a 2,000-year-old scroll that had been rendered unreadable after being found burnt from the eruption of Mount Vesuvius in Pompeii in 79AD.
Do Green AI Initiatives Work?
While individual users can take small steps to lower their carbon footprints, real change can only happen if the major companies that use and develop AI make changes to their policies. A 2024 study in the Journal of Cleaner Production on green artificial intelligence initiatives dives deep into the significant energy consumption of AI. The study looks at ethical considerations, open-source initiatives and collaborations, and other factors surrounding the surge of AI. This study suggests 55 initiatives to optimize the cloud and improve the efficiency of the AI model to significantly lower the negative environmental impact.
So far, over 190 countries worldwide have adopted non-binding recommendations by the UN Environmental Programme regarding the ethics of AI. Even though the environmental impact is part of the blanket agreement to use AI responsibly, there are very few actual policies or legislations specific to protecting the planet or lowering emissions. Since the boom of AI is so new, a lot more research is needed to evaluate the full extent of artificial intelligence and how to create real change.
Opinion: As An Individual User, How Can I Help?
Ever since I learned about the environmental impact of AI and machine learning, I have looked into ways I can make a difference. As with any global environmental concern, everyone can take small steps to decrease their AI carbon footprint. While I’m concerned, I am also curious about AI and all it can offer me. My personal approach is to be mindful whenever I use artificial intelligence. Whenever I can, I use eco-friendly alternative search engines, like Ecosia, which plants trees to offset their carbon emissions.
As an iPhone user, I don’t see myself turning off Apple Intelligence completely, but I have toggled off features I won’t ever use, such as notification summaries. Plus, I try not to unnecessarily play with any AI image or video generators and only use generative AI when I truly need it.
I think the most important thing we as individuals can do is to learn more about artificial intelligence and push companies and governments to do more research and be as transparent as possible. AI is here to stay, and I believe we can learn how to use it responsibly and safely for everyone concerned.
Top image credits: Syahrul38 / Shutterstock.com
Naiswari Graphics / Shutterstock.com
Blan-k / Shutterstock.com
Serhiy Smirnov / Shutterstock.com
