Artificial Intelligence or AI has transformed our lives in the most profound ways. It has become a great ally in the battle against climate change. However, AI has a secret: it also contributes to greenhouse gas emissions. So – how can AI become greener?
How to make AI greener
Before tackling how AI can be made greener, specific facts need to be established.
One, creating and using AI systems incur huge tangible costs. For example, OpenAIs language model GPT-3 consumes energy with a carbon footprint equivalent to driving a car from the Earth to the moon and vice versa.
However, its numerous beneficial impacts on climate change cannot be discounted. A 2020 study was undertaken to evaluate AIs potential impact on the UNs 17 Sustainable Development Goals.
Results from that study indicated that AI can potentially facilitate 93% of identified environmental targets, such as Internet-of-Things devices, smart and low-carbon cities, smart grids to better integrate renewable energy, combating marine pollution, and satellite imagery to identify desertification trends.
Case studies to learn from
Several case studies can give us an idea of how AI can help the fight to reduce greenhouse gas emissions. A case in point is how a cement firm and AI collaborated to lower its overall energy consumption.
Turkish cement firm OYAK Cimento has utilized AI to lower the companys carbon footprint significantly. Berkan Fidan, OYAK Cimentos Director on Performance and Process, AI-assisted processes implemented at an enterprise level have resulted in greater operational efficiency, higher production gains, and ultimately, lower energy consumed.
A medium-sized cement plant capable of producing 1 million tons of cement that uses AI-powered quality control mechanisms can reduce its carbon emissions by 7,000 tons annually. This is equivalent to the carbon absorption of 320,000 trees.
Another illustration of the potential environmental impacts of AI is with Chilean telco Entel. The firm is utilizing sensor data for the identification of forest fires. Given that Chile is a highly climate-vulnerable country, it suffered a catastrophic wildfire in 2017 that destroyed 714,000 acres of forestland.
Entels Digital Unit, Entel Ocean, used IoT sensors to locate forest fires. By placing these sensors on trees and having these sensors smell air particles, AI helps Entel predict where the next forest fire is likely to occur. The firms efforts have allowed it to detect a forest fire 12 minutes earlier than conventional methods.
By these case illustrations, its clear that AI can be a powerful weapon in the fight against climate change. Nevertheless, there are trade-offs because its also a contributor to carbon emissions.
So how to strike a balance?
First, making AI greener requires a multidimensional and more holistic model evaluation. The default mindset when it comes to AI is that bigger is better. However, this thinking needs to be rethought especially in placing AI within the environmental space.