One of the main topics of today’s society is climate change and how to stop it. Every organization is investing in greener and more sustainable operations. We all know the basic solutions like electric cars and solar panels, but did you know that AI can play a role in fighting climate change as well?
AI can affect the climate positively. AI is applied in multiple different industries and as a result it has multiple purposes. For example, a big environmental organization like WNF uses AI to combat deforestation. Moreover, NASA and IBM are working together on a new climate model that can predict weather and climate change with the use of AI.
But unfortunately, the effects of AI on the climate are not only positive. AI also has some negative impact. In this blog we’ll delve into the negative and positive effects of AI on our environment. Additionally we’ll look at how AI is getting more sustainable.
Key Learnings
- The training of an average AI model produces as much CO2 as driving 62.6 cars for a year.
- The training of the GPT-3 model has consumed 700.000 liters of water.
- AI has the potential to solve a lot of climate problems, like deforestation.
- With the help of AI, organizations in multiple industries, already make their operations more efficient and reduce their carbon footprint.
- AI is becoming more sustainable with new technologies supported by the Green AI movement.
The rapid rise in popularity of AI
AI is booming and organizations realize that it’s important to jump on the bandwagon now to survive. The adoption of AI experienced exponential growth in recent years. Organizations are looking for new opportunities to innovate. Moreover, the need for digital transformation was significantly sped up by the covid pandemic
Why climate impact of AI is important to talk about
Talking about the impact of AI on our climate is becoming more important as it’s playing a bigger role in organizations. This increasing use results in more resources being needed to keep up with demand, and those resources have impactful effects on our climate.
But, we should not only look at it in a negative way, research and projects have shown it can also be a helpful asset to combat climate change. Potentially AI’s impact could help solve many climate issues.
The environmental impact of AI
As we talked about AI impacts the climate both negatively and positively. The positive effects can’t exist without the negative impact. So the goal is to find a balance where we minimize the negative impact and maximize the positive effects. We’ll first look at the negative impacts.
Negative impact: Energy Consumption & CO2 Emission
The negative impact of AI is mostly related to the data centers. These data centers are large and energy slurping buildings that are used to keep AI models up and running.
Compared to typical commercial buildings, data centers consume 10 to 50 times the energy per floor space. Energy consumption of these data centers is still on the rise due to Large Language Models (LLM), like GPT-4, becoming bigger in size.
It’s hard to determine the exact carbon footprint of AI. Big tech companies don’t like to share their exact emissions and energy usage. Also, there is no predefined way of measuring emissions related to AI. We do have research that has tried to capture the carbon footprint related to AI.
Some statistics that give an indication of AI’s negative impact on the climate:
- According to the US office of Energy Efficiency and Renewable Energy, all data centers together account for approximately 2% of the total electricity use in the United States. And AI uses a significant part of the data centers capabilities.
- NVIDIA is expected to ship 1.5 million AI servers per year by 2027. These servers will consume at least 85.4 terawatt-hours of electricity per year. To put these numbers in perspective: in 2021, residential households from Switzerland consumed 19.4 terawatt hours of electricity.
- One recent study states that ChatGPT consumes 500ml of water for every simple 20-50 questions and answers. The water is used to keep data centers cool. The used water can contain chemicals and other waste, so in order to reuse it, it needs additional processing.
- The same study also states that training GPT-3 could potentially have used up 700.000 liters of clean fresh water to prevent the data centers from overheating. This water is evaporated, so not reusable, which increases water shortage.
- A study from the University of Massachusetts found out that training one AI model can emit more than 283.000 KG (626.000 pounds) of carbon dioxide. These emissions are comparable to driving 62.6 average gas powered vehicles for a year.
- Generating an image with Generative AI, like OpenAI’s DALL-E 3, costs the same amount of energy as fully charging your phone.
To get a better understanding of AI’s carbon footprint, we need more data from the organizations that build the models. For example if they use renewable energy or fossil fuels. For companies it’s not required to share information about energy consumption of AI.
In order to get more information, we need law changes. These law changes will probably come up in the near future. There are no AI laws yet, but the EU recently passed the new EU AI Act. The act does mention that AI companies should build sustainable and environment friendly models and that they should be transparent about their AI systems. However, the act doesn’t mention specific laws or guidelines regarding this issue.
Positive impact: Green AI
In the world of AI there is a movement called Green AI. The goal of Green AI is to design and develop algorithms that are sustainable and environment friendly. The movement wants to optimize energy in the most efficient way, reduce greenhouse gas emissions, and promote sustainability practices.
One important feature of AI is that it makes it easier to sift through large amounts of data, for example from a weather satellite. This makes it possible to monitor climate change faster and on a much larger scale. It can also help with natural disasters with regards to accurate prediction, efficient evacuation, and damage mitigation.
Positive Impact: Preserve Nature
AI has a great potential to help against climate change. Especially when it comes to preserving nature it has several applications.
- Combat Deforestation: In the beginning of the blog it was already mentioned that WWF uses AI to combat deforestation. WWF’s AI model Forest Foresight analyzes big data like satellite images to detect and prevent illegal deforestation in an early stage. For example, it detects new roads that could be used for the transportation of woodcutting machinery.
- Fight Wildfire: AI is used to fight wildfire. By collecting and analyzing data like camera footage and 911 calls, the computer model can locate and predict the growth and direction of the fire. This AI tool allows firefighters to act more accurately and quickly.
- Detecting Wildlife Diseases: AI can be used to detect and manage diseases of wildlife in an early stage. AI algorithms are used to scan digital images and detect signs of diseases. When compared to humans, AI is much faster and isn’t prone to human errors. With this AI tool the disease can be dealt with in an early stage and prevent it from spreading.
- Track Biodiversity: Scientists are combining AI algorithms with drones and satellite footage to monitor biodiversity. The AI tools can analyze and recognize changes in animal behavior and track animal populations and their migration. The tool is also used to track the movement of marine life and to identify changes in coral reefs, so scientists can act quicker.
- Prevent Overfishing: The Smart Boat Initiative of the Environmental Defense Fund (EDF) has several features to help fight overfishing. For example, they use AI tools to recognize fish species and measure size. This tool helps fishermen in their sorting process and prevents them from taking too much or the wrong fish. Furthermore, OceanMind has built an AI tool that tracks down fish boats, which makes it easier to track down illegal fishing practices.
Positive Impact: Circular Economy
A circular economy means that the value of the products like labor, energy, and materials are preserved. It focuses on durability, reuse, remanufacturing, and recycling to keep products, components, and materials circulating in the economy. AI can help with this.
Identifying and recovering recyclables: AI is more capable of identifying and recovering recyclables compared to humans. Companies like AMP Robotics and MachineX have developed AI-guided robots that recycle. Compared to humans, these robots are two times faster and more consistent in picking up recycled materials. AMP Robotics claims that the technology already helped avoid approximately 1.8 millions metric tons of greenhouse gas emissions, which equals the emission of almost 375.000 cars.
Recycle and reuse water: AI also makes it easier to recycle and reuse water, by monitoring water quality. It identifies contaminants and signs of pollution. With AI, water pollution will be detected at an early stage so water agencies can act on time. When the water is already contaminated AI can be used to detect and extract the chemicals.
Positive Impact: Optimization of Operations
This blog already stated that AI consumes a lot of energy. But it also saves a lot of energy. Some organizations apply AI to optimize their operations, increase sustainability, and reduce their carbon footprint. AI is applied in a wide variety of industries including agriculture, transportation, and manufacturing.
Agriculture: AI is known for its ability to process a lot of data in the blink of an eye. And that’s exactly why it helps enhance precision agriculture. By using data on weather, soil, and crops AI tools can increase yield, reduce environmental harm like water waste, and promote sustainable farming methods.
Transportation: AI can also help with optimizing transportation routes. Generating the most efficient routes will lead to less gas emission from vehicles. Additionally, AI can be used to solve traffic flow problems. By analyzing traffic and predicting potential congestion, traffic jams can be avoided or even prevented, which also leads to less emissions.
Waste Management: Another important topic in the climate debate is waste management. AI can help with dynamic pricing on food that’s almost expired. It can also improve inventory management and prevent waste by predicting customer behavior. Furthermore, AI is able to predict maintenance, which prevents expensive repairs and increases sustainability.
AI can help organizations with building sustainable supply chains. With predictive algorithms and efficient production and transportation, AI can reduce the carbon footprint of organizations.
How AI becomes more sustainable
Like every other industry, the AI sector tries to become more sustainable. New technologies will allow AI to emit less CO2, make e-waste reusable and make way for more applications in fighting climate change.
Green Data Centers
There’s still much to be gained in the energy consumption of data centers. The concerns about climate change are rising amongst organizations. This led to the emergence of so-called green data centers.
The green data centers use renewable energy and because of that they’re able to reduce energy consumption, carbon emissions, and costs. A large tech company like Google is already carbon-neutral because they use renewable energy and optimize energy consumption with machine learning.
Reuse of water
Not only does the energy consumption of data centers form a problem, also the extensive use and pollution of water is an issue.
New techniques make it possible to reuse the high temperature water that are produced by data centers. According to Ramboll, data centers can deliver excess heat to a heating network. This heated water can be used to heat buildings, greenhouses, and pools.
Underperforming AI models
By shutting down underperforming AI models in an early stage, the energy used for the training of AI models could be reduced by 80%. This could contribute a lot to making AI more sustainable.
What does Klippa and DataNorth do to reduce its environmental impact?
At Klippa and DataNorth we are also investing in sustainability. Most importantly, we plant trees to compensate our emissions. In 2022 we planted more than 1000 trees via ‘Trees for All’ and compensated for 630 tons of CO2. Additionally, we help organizations with their digital transformation that makes paper documents redundant.
Do you want to know what kind of positive impact AI can have for your business, with the help of DataNorth? Please contact us here!
