Margaret Korte
Print/Podcast Editor
As AI usage becomes more widespread, its ethics also become more controversial.
Opinions differ on AI’s role, whether it is a tool or a crutch, a step towards a more innovative future, or a dystopian one. Its convenience has been both praised and trashed–it makes life easier but may hinder one’s ability to think critically.
Recently, studies and articles have sought to finally end this discourse by arguing that AI has harmful impacts on the environment, settling the argument once and for all.
Research on AI is still scarce, and little can be definitively known; however, the research points largely to its harmful effects.
What is known is that AI uses a lot of energy and electricity. The training of an AI tool requires a “staggering” amount of electricity, according to an article published by the Massachusetts Institute of Technology (MIT). This “leads to increased carbon dioxide emissions and pressures on the electric grid.”
Because AI technology continues to advance, so does its training, and each model requires more energy.
For example, scientists have estimated that the training for GPT-3 by OpenAI took 1287 megawatt hours of electricity, which is “enough to power about 120 U.S. homes for a year,” releasing about 552 tons of carbon dioxide.
The AI training process also requires “rapid fluctuations in energy use,” which power grids must accommodate for, usually using diesel-based generators.
The hardware used to train AI is also cooled with water. According to MIT, excessive water use for this purpose “can strain municipal water supplies and disrupt local ecosystems.”
Data centers, where AI models are trained, are not new: Amazon has over 100, but they have “dramatically increased” since the rise of AI.
The electricity consumption of data centers was at 460 terawatt-hours in 2022, making it the 11th largest electricity consumer, placing it “between the nations of Saudi Arabia and France, according to the Organization for Economic Co-operation and Development,” according to MIT.
It is estimated that by 2026, electricity consumption of data centers is expected to approach 1050 terawatt-hours, placing it at number five, between Japan and Russia.
The demand for data centers also “cannot be met in a sustainable way,” according to MIT. “The pace at which companies are building new data centers means the bulk of the electricity to power them must come from fossil-fuel-based power plants.”
“Pushing up demand for fossil fuels means more CO2 is emitted, which helps global temperatures creep up,” Kenza Bryan, climate reporter, told PBS in an article.
According to the MIT article, researchers also estimate that “a ChatGPT query consumes about five times more electricity than a simple web search.” Other sources have argued that it’s no more electricity than is used by watching a few hours of Netflix. The impacts of AI, then, may be harmful, but no more than any other internet activity.
Despite these arguments that AI is causing negative impacts, some sources have argued that the lack of research means it can’t be stated for certain.
According to Yale Climate Connections, “Spotty, unreliable, and missing data make it incredibly hard to accurately assess AI’s true climate impact and energy needs, let alone figure out what to do about it.”
Until research becomes more reliable, AI users should remain cautious, using shorter prompts to cut back on the amount of energy used or limiting AI use altogether.
AI users should stop and ask themselves, “Do I really need ChatGPT for this?”
Try shifting the mindset. If, when confronted with an issue, one’s instinct is to think, “Let me ask ChatGPT,” a change needs to be made.
Regardless of AI’s environmental impacts, taking a moment to think or talk over an issue, to read or write something thoroughly, will always be beneficial.
 |
Margaret Korte–Print/Podcast EditorThis will be Margaret Korte’s third year on ECHO staff. She made several contributions while taking journalism class her freshman year. |
Support our Sponsors
