How Artificial Intelligence Helps the Environment & Sustainability

The problems facing our planet these days can seem overwhelming. The systems our modern society depends on—agriculture, industry, transportation—are polluting our air, water, and soil. Global warming is causing devastating heat waves, wildfires, droughts, and storms. Forests are disappearing, coral reefs are dying, and species all over the world are threatened. All this makes it harder than ever to provide water, energy, and food for our growing population.

But humanity also has a powerful new tool for dealing with these challenges: artificial intelligence, or AI. It can collect and analyze environmental data much faster than human scientists using older tools. And it can use this information to predict problems and find the best solutions.

The role of AI in environmental solutions

The term “artificial intelligence” is a bit misleading. The most sophisticated machines we have today can’t really think or make decisions on their own (...yet). However, they can perform a variety of functions that would once have required a human brain. They can collect and analyze data, make predictions, and use language. They can also learn to do all these tasks better the more they practice.

Modern AI isn’t a replacement for human intelligence but a supplement to it. It can make humans much better at:

• Monitoring environmental problems. These days, it’s possible to gather vast amounts of environmental data remotely using sensors, satellites, and drones. AI can use these sources to measure problems like pollution and deforestation and target efforts to remediate them. It can also track weather emergencies, such as storms.
• Optimizing agricultural practices. AI can play a key role in making farming more sustainable. Algorithms can help farmers determine the best times to plant, spray, water, fertilize, and harvest their crops. This allows them to maximize yields while minimizing their use of water and agricultural chemicals that can damage the environment.
• Modeling climate change. Scientists use AI to help them understand climate trends and how our actions shift them. For instance, AI tools can predict how global warming affects the frequency, intensity, and lingering effects of natural disasters. This helps people and governments figure out which areas are at risk and how to prepare.
• Reducing emissions from transportation. AI can make transportation more efficient in a variety of ways. It can optimize the fuel efficiency of autonomous or AI-assisted vehicles. It can improve public transportation routes to minimize the number of vehicles on the road. And it can streamline traffic flow by helping drivers avoid traffic jams and other obstacles, minimizing driving time.
• Using energy efficiently. Smart sensors and meters can boost the energy efficiency of buildings by delivering energy only where it’s needed. Google has done this to reduce cooling costs in its data centers, cutting their energy use by 15%. AI can also improve the efficiency of utilities and the power grid, reducing wasted energy. The Information Technology Industry Council estimates that “intelligent efficiency” measures could cut total U.S. energy use by 12% to 22%.
• Delivering clean drinking water. AI-powered systems can monitor water quality and track usage in municipal water systems. This helps the utilities find leaks and determine when maintenance is required. Water scientists can also use AI to predict water levels in an area and determine when restrictions are needed.

AI-powered environmental monitoring

Artificial intelligence helps scientists gather and analyze all kinds of information about the environment. AI tools can help them monitor:

• Air quality. AI can use data from sensors, satellites, and weather stations to monitor air quality in real time. It can detect air pollution, identify its source, and target efforts to remove or reduce it. For example, the United Nations’ GEMS/Air Program uses AI to analyze air quality data from over 25,000 monitoring stations worldwide. This helps government officials take appropriate steps to protect public health.
• Water quality. Sensors and drones paired with AI can track water quality measures such as temperature, acidity, and chemical contamination. They can identify plastic pollution, coral bleaching, illegal fishing, thermal pollution, and disease. After spotting these problems, they can help scientists identify their sources and figure out how to address them. AI can even use predictive modeling to determine the chances of future problems, such as algae blooms or extreme weather.
• Weather patterns. AI tools can monitor natural disasters in real time—both before and after they strike. This capability has many benefits. In some cases, it can stop disasters before they happen. For example, a tool called Silvanet uses AI-enabled sensors to detect wildfires while they’re small and easier to contain. It can also warn of impending weather emergencies. A startup called Climavision is using AI to improve weather predictions, and a Japanese “smart city” offers AI-enabled disaster alerts to its citizens. Lastly, AI can analyze the aftermath of a disaster to predict its long-term effects so individuals and governments can deal with them.

Predictive analytics and climate modeling

One reason climate change is such a difficult problem to deal with is that it’s hard to predict. However, artificial intelligence can make it easier. AI tools analyze historical climate and pollution data and also monitor current greenhouse gas levels, glacier mass, and sea level. For instance, Carbon Tracker combines AI with satellite imagery to track greenhouse gas emissions from fossil fuel power plants. AI can then process the vast amounts of data involved to model how all these factors are connected. This helps scientists determine how climate change affects ecosystems and human populations.

One of the most ambitious AI-based climate models is the European Commission’s Destination Earth (DestinE) project. This digital “twin” of the earth will model the changing climate as accurately as possible over time. In addition to mapping the planet’s past, DestinE will run simulations showing how it could change in the future. Governments will be able to use this information to develop better policies for both mitigating and adapting to climate change.

Natural resource management

Artificial intelligence can also be useful in managing natural resources, such as forests, fisheries, and bodies of water. AI can analyze features of a landscape, such as plant cover and water flows, and detect problems like soil erosion. This helps people in the area address the problem and prevent further loss of natural resources. AI can also help us restore damaged landscapes, such as forests or coral reefs. It can identify the ideal variety of new species to add and the best places to put them.

One area in which AI is particularly helpful is forest management. For instance, a company called NCX has used AI to map out every acre of forest in the continental U.S. Having an accurate inventory helps landowners and foresters know which trees to cut and which to leave behind. AI tools also help forest managers perform “controlled burns” to remove excess brush that can fuel wildfires. Using AI, they can determine when weather conditions are ideal to set a fire without endangering nearby communities.

Biodiversity conservation and wildlife protection

Another type of natural asset artificial intelligence can help manage is wildlife. Scientists can use AI programs to:

• Monitor the health of plants.
• Track the population of endangered species.
• Predict the migration patterns of birds.
• Detect the presence of invasive species.
• Detect poaching (unauthorized hunting or trapping) using drone images.
• Predict the routes used by poachers so rangers know where to patrol.

One example of an AI-driven platform used for conservation is Wildbook. It uses machine learning and computer vision to recognize and track the movements of individual animals based on photos and other data. This helps scientists keep track of population and migration patterns and develop strategies for conservation.

Another AI tool for protecting wildlife is Protection Assistant for Wildlife Security (PAWS). It employs machine learning to predict the behavior of poachers. Based on this information, it creates “risk maps” of an area and suggests patrol routes for rangers. In its first month of testing, PAWS enabled rangers to remove twice as many illegal snare traps as usual. The researchers behind PAWS are also working on an AI tool to map smuggling of animal parts. This will help authorities position checkpoints to catch illegal wildlife traders.

Environmental remediation and clean technologies

Two of the biggest environmental problems are plastic waste and emissions from burning fossil fuels. Artificial intelligence offers several ways to address these problems. It can assist with:

• Air pollution control. AI can limit air pollution in a variety of different ways. As noted above, it can reduce vehicle emissions by analyzing and optimizing traffic flow. It can also analyze industrial activity and propose measures for cutting emissions. Also, companies like Climeworks use smart algorithms to improve the efficiency and cost-effectiveness of carbon capture.
• Sustainable energy. AI is also helping to power the transition from fossil fuels to clean energy. For starters, it can optimize the performance of wind turbines and solar panels. Wind farms using Google’s DeepMind AI have boosted their output by 20%. AI is also a key component of smart grids that streamline the delivery of clean energy to users.
• Energy efficiency and conservation. Another way to reduce fossil fuel use is to use less energy overall. AI can help with this by improving the efficiency of buildings, vehicles, and industrial processes. For instance, smart home technology can save energy in buildings by automatically turning off lights or adjusting thermostat settings.
• Waste management. AI can reduce landfill waste by helping waste management facilities sort and process recyclable materials better. Companies like AMP Robotics use AI-equipped robots to identify different materials and recover the recyclable ones. AI can also improve waste collection by optimizing collection routes and schedules, reducing emissions from garbage trucks.
• Ocean cleanup. Along with reducing future pollution, AI is helping to clean up pollution that already exists. One notable example is the Ocean Cleanup Project. It uses ships equipped with sensors and machine learning systems to locate and remove plastic waste from the ocean.

Challenges and ethical considerations

Artificial intelligence is a powerful tool for creating a cleaner environment. However, this revolutionary technology has great potential to be harmful as well as helpful. There are three major ethical concerns related to AI in environmental solutions: privacy, bias, and resource use.

Privacy

One problem associated with AI is privacy. Training an AI requires huge amounts of data. If an AI learns from a data set that includes people’s personal information, what does it do with that information? Could anyone who uses the AI gain access to it? The answers aren’t always clear.

There’s even a risk that an AI could infer information it wasn’t actually given. It might be able to deduce an individual’s location or habits based on general data about that person.

One example of how this could play out in environmental applications relates to energy use. AI programs can help you save energy by keeping a record of your energy use at home or at work. But based on this data, the AI might be able to figure out details about your schedule and habits. It might even be able to identify you by name.

To deal with this problem, businesses and other organizations are working on ethical guidelines for training AI. Examples include the Partnership on AI, the United Nations' Multistakeholder Advisory Body on Artificial Intelligence, and tech companies like IBM.

Bias

You might think that bias is a human problem. Computers are machines that operate according to logical rules, so their judgments ought to be strictly impartial. But just like humans, AIs learn what they’re taught. If the data set used to train an AI is skewed in some way, the AI’s output will be similarly skewed or outright wrong.

Experts say countering bias in AI is a two-part process. The first step is to come up with ways of defining and measuring fairness. The second is to incorporate that definition into an AI’s decisions. This could be done either by making it part of the AI’s training or by using it to adjust the system’s output.

Resource use

Ironically, the AI tools being used to help the environment also create environmental problems of their own. AIs that crunch huge amounts of data use vast amounts of computing power. The computers that run them consume lots of energy and require lots of water to keep them cool. Already, data centers use around 1% of all the world’s electricity and produce around 1% of total greenhouse gas emissions. By 2027, AI servers could consume as much energy as a small country.

Another problem is electronic waste, or e-waste, from the computers that run AI programs. This waste often contains toxic materials such as chromium and lead. Also, it requires rare materials, and generates e-waste. In 2021, the world’s e-waste totaled over 63 million tons—more than the mass of the Great Wall of China. This number will only grow as AI becomes more prevalent. And currently, only around 17% of e-waste is properly recycled.

The United Nations’ Coalition for Digital Environmental Sustainability (CODES) is working on ways to reduce AI’s environmental footprint. Changing the design of AI programs can reduce their energy use almost 90%. Solutions for reducing e-waste include making it easier to repair and recycle electronics.

The potential of AI for protecting the environment

Artificial intelligence is a fairly new technology, but it’s already transforming our approach to protecting our planet. AI helps us monitor the environment—air, water, wildlife—to detect and respond to problems. It can also create models to help us identify the best solutions.

In the future, AI could do even more. Artificial intelligence is growing more intelligent all the time, able to handle even bigger and more complicated sets of data. And the growing Internet of Things—physical devices linked to the Internet—will provide more data for it to use. This will make AI’s predictions more accurate and its solutions more useful.

The problems facing the planet are large and complex. But thanks to AI, we now have large and complex tools for tackling them.