Considerations for AI and the Climate Crisis
Artificial intelligence is an umbrella term for software systems capable of making decisions that traditionally would require a human brain. Machine learning—typically used by researchers—is a subset of AI designed to understand existing datasets, whereas generative AI is capable of creating new content and has become popular with the general population.
AI requires a lot of computer processing, which means the AI boom has greatly increased the energy demand of data centers—giant facilities filled with computers that make things like AI and online marketplaces possible—as well as put pressure on fresh water supplies, because it takes a lot of water to keep data centers from overheating.
The skyrocketing environmental footprint of AI is largely attributable to the mass-consumption of generative AI for tasks that range from unnecessary—generating bizarre images or fake videos—to those that can only be described as ‘isn’t this why the human brain exists?’—like homework, college assignments, or work reports. The main purveyor of such redundant applications of generative AI, ChatGPT, uses as much electricity in one day as 33,000 U.S. households use in a year.
The carbon-intensiveness of generative AI deserves to be accentuated, instead of avoided, in conversations about AI. And we must also recognize that machine learning-type AI is a useful tool for scientific research, including understanding and adapting to climate change.
Machine learning systems can rapidly analyze vast amounts of data, identify patterns, and make decisions and predictions. The ability to process massive amounts of data has useful research applications. For example, machine learning tools can be trained to quickly identify and locate humpback whale songs in 180,000 hours of underwater recordings—a task that would take a single researcher 20 years to do manually. They can also efficiently analyze vast amounts of conservation data, enabling conservationists to quickly respond to a world that is being rapidly altered by climate change. Machine learning tools can also be applied in uses as varied as management of aquaculture and fisheries and optimization of offshore wind energy. And AI-powered autonomous underwater robots have become vital tools for exploring the deep sea, tracking changes in ocean chemistry, and mapping the seafloor.
The question remains, with the global imperative of reducing global emissions in order to limit global warming to 1.5 degrees Celsius above pre-industrial levels, is it wise to be expanding a novel carbon-intensive industry? For those concerned about climate change, the answer should be clear—let’s stick to the utilitarian uses of AI, and skip the convenient but unnecessary applications. This is the newest environmentally-conscious lifestyle choice: the bring-your-own-bag movement of the internet age.
Written by Emily Vidovich. Emily is an environmental journalist specializing in ocean conservation and climate change mitigation. She obtained her bachelor’s degree at George Washington University and a Masters in Global Environmental Studies at a university in Tokyo, Japan. Born and raised in the Port of Los Angeles, she now works in research and communications at AltaSea.