The technological advancements of Artificial Intelligence (AI) are revolutionary, but burns copious amounts of energy in its training and usage — far more than traditional technology. AI’s excessive energy use when created and operated, along with its immense e-waste production, must end as it accelerates environmental harm that can destroy the planet.
AI training sessions both teach AI models to perform tasks by analyzing data and disproportionately affect the planet. Training AI is an extremely energy-intensive process, a 2024 study by Cornell University estimated to be around 33 times the amount task-specific software uses. Before it is even operated, it uses alarming amounts of energy surpassing previous technologies. Vrije University Amsterdam researcher Alix de Vieres found in his study “The growing energy footprint of artificial intelligence,” training models such as GPT-3 burn around 1,300 MWh of electricity — roughly the annual power consumption of 130 US homes. Furthermore, a study at the University of Massachusetts cited by Earth.org also determined that an individual AI training session produces up to 636,000 pounds of carbon dioxide, equivalent to the emissions in 300 round-trip flights between New York and San Francisco. AI training sessions are only the beginning of AI’s environmental footprint.
Meanwhile, operating AI consumes enormous amounts of energy as its systems are constantly running. According to de Vieres’s study, each use of AI services cost 10 times the electricity of a standard keyword search (at around 3 Wh per use). If Google were to continue its path in implementing AI in every search through its Gemini model, the electricity powering this AI coil total 2.93 TWh — Ireland’s annual energy usage.
This excessive energy consumption releases carbon dioxide and other greenhouse gasses into the atmosphere, making up 2-3% of global emissions, according to findings from the UN’s Environment Programme. As Linköping University researcher Anders Nordgen’s published in his study, “Artificial intelligence and climate change: ethical issues,” AI emissions are expected to reach 14% of global emissions by 2040. The emissions AI creates are a glaring concern that must be addressed before they completely dismantle the planet and undermine the safety and health of the very people AI is meant to serve.
In a world trying to prioritize sustainability, the production of AI drives the opposite.
Along with the exacerbation of energy, AI usage also strains resources. As it requires specialized machinery and hardware such as powerful processors and GPUs (graphics processing units) to run, as demand increases, new models are developed, rendering older ones obsolete and discarded to landfills. As found in a study by the World Economic Forum, cited by Earth.org, by 2050, the amount of e-waste generated is projected to surpass 120 million metric tonnes. Additionally, as Renée Cho from the Columbia Climate University states, AI equipment like GPUs are made of rare elements such as lithium and cobalt, only obtainable through mining that causes deforestation, destroys ecosystems and displaces wildlife. Creating AI is harmful for environmental sustainability.
The AI revolution has yielded positive impacts, with many using it as a tool to increase efficiency. Farmers have implemented AI to determine more effective farming practices. However, it prioritizes crop yield maximization over sustainability since it lacks human insight and ethics, often opting for extreme pesticide use. AI’s environmental impacts are still rampant, with choices harming biodiversity.
Several approaches must be considered to mitigate AI’s environmental impact. Developers must prioritize using more energy-efficient methods and extend the lives of AI equipment. Additionally, it is crucial data centers transition to renewable energy sources, reducing AI’s immense energy consumption and environmental footprint. Efforts must expand from Google and Microsoft’s investment into renewable solar or wind-powered energy to diminish the detrimental environmental effects caused by AI’s energy consumption.
AI’s environmental impacts are catastrophic for the future of earth and its inhabitants. If demand for AI and its effects are not addressed, the resulting harm will impact access to water, food, health, and the overall quality of life for future generations. In a world trying to prioritize sustainability, the production of AI drives the opposite. Individuals must hold corporations accountable for how AI is developed, as society works together to find sustainable solutions to ensure that this technology serves both humanity and the planet in a responsible manner.