By Alessandro El Khoury | Staff Writer
Since its inception, artificial intelligence has been invading every aspect of our lives. It can recognize faces, detect fraud in the banking system and diagnose disease. In addition to this, AI can produce at an unprecedented pace. What takes us days to process, AI can deal with in seconds. Nonetheless, it would be fair to ask about the impact of such technology on the environment. Indeed, as AI evolves it is possible that it might require an increasing amount of energy.
One of the environmental costs of AI is greenhouse emissions. By 2040, Information and Communications Technology will contribute to 14% of global emissions. Specifically, researchers at the University of Massachusetts reported that certain AI models produce 5 times the lifetime emissions of a car.
Furthermore, another environmental concern is E-waste. AI consists of hazardous chemicals such as mercury which can contaminate soil and water. The World Economic Forum predicts that by 2050, the E-waste burden will reach 120 million metric tons.
AI may also endanger natural ecosystems. If AI is applied to fields such as agriculture, it may result in greater use of pesticides and fertilizers which pollute soil and water. Another challenge to overcome is biased AI. Sometimes the design of a model may overlook certain aspects or be based on a sample that is not representative of the general situation. This results in AI models that are biased. An AI model could be biased into favoring economic growth over environmental protection and thus harm the ecosystem.
Despite these burdens, AI can work in the opposite direction. According to Jérôme Stubler, CEO of Equans: “Using artificial intelligence, we can gradually understand the thermal inertia of a building.” Since AI can handle considerable amounts of data, it can optimize energy management at the urban level as the demand of each building requires. This will lead to a decrease in CO2 emissions. According to Equans, AI can limit the emissions of transportation processes through efficient planning. For instance, Google’s AI-focused DeepMind initiative reduced fuel consumption in vehicles by 15% by limiting commuting times with the help of sensors which record traffic data.
According to the UN, AI can build models which can predict climate change trends and accordingly set appropriate strategies. AI can also limit pollution. In fact, it can organize traffic and waste management. AI can also predict climate disasters. For instance, areas susceptible to landslides, mapping can assist governments in establishing the appropriate development strategies to keep the environment as well as its people safe.
The impact of AI is not clear cut. Any AI process has drawbacks. It consumes much more energy than other infrastructures and leaves more waste behind it. If it is biased it can turn one problem into multiple ones. On the other hand, it can sometimes limit energy consumption in certain sectors and predict danger before it reaches us. The key to the issue will be balancing advantages with disadvantages with the following principle: favoring AI models with minimal negative effects on the environment and maximal protective effects when it comes to the environment. As such, this draws light on the importance of establishing a well-recognized policy at the global scale.
Sources:
Kanungo, Alokya. “The Real Environmental Impact of AI.” Earth.Org, 17 July 2023, earth.org/the-green-dilemma-can-ai-fulfil-its-potential-without-harming-the-environment/.
“Explainer: How Ai Helps Combat Climate Change | UN News.” United Nations, United Nations, 2023, news.un.org/en/story/2023/11/1143187#:~:text=Machine%20learning%20can%20optimize%20supply,efficiency%20across%20energy%2Dintensive%20sectors.
“How Artificial Intelligence (AI) Is Driving Decarbonisation.” Austria, www.equans.com/news/how-artificial-intelligence-is-driving-decarbonisation#:~:text=By%20leveraging%20machine%20learning%2C%20AI,resources%20and%20reduced%20carbon%20emissions. Accessed 5 Jan. 2024.