It takes time, quality data, and a lot of electricity to train artificial intelligence. It took about 27,648 kilowatt-hours (kWh) for the AI research and implementation company OpenAI to train the GPT-3 language model for 9 days. Considering that one US household consumes 10,715 kilowatt-hours (kWh) per year, the cost of AI is enormous. If this technology is so energy-consuming, how can it help achieve environmental sustainability? Let’s consider several cases of how AI improves the environment.
The United Nations has identified 17 sustainable development goals, which can be divided into three groups – “Environment,” “Economy,” and “Society.” AI can directly or indirectly accelerate the solution of every problem, from protecting endangered plant species to reducing carbon dioxide emissions into the atmosphere.
AI and climate improvement
For many years now, humanity has been struggling with climate change and global warming caused by an increase in the concentration of greenhouse gases in the atmosphere. Electricity and heat production, agriculture, and transport remain the “leaders” in terms of emissions of carbon dioxide, nitrogen oxide, methane, and other substances. Gases absorb infrared radiation and retain heat, causing a gradual increase in temperature.
In its 2021 study, the Intergovernmental Panel on Climate Change (IPCC) warns that the global carbon balance will run out by 2030. Going beyond this limit will accelerate global warming by more than 1.5 degrees. There is no time for reflection, and humanity seeks to smooth the results of its activities by attracting technologies for these purposes.
AI is one of the options for slowing down global warming. The independent think tank Carbon Tracker taught a smart algorithm to monitor power plant emissions from satellite images. Based on the collected data, the research organization makes predictions about which power plants hide the real extent of pollution. It also determines which ones are worth investing in to ensure a low-carbon future.
If AI is properly trained, the algorithm will predict the demand and supply of electricity. The technology will detect the possible risks associated with the imbalance of supply and demand and the production and use of electricity in advance. EPRI collects data to teach AI how to solve complex problems in the energy industry. For example, how to create an application that will improve refinery operations and mitigate environmental impacts.
- Source: nature.com
AI to protect plants and animals from extinction
According to the UN, about 1 million species of plants and animals are endangered due to the negative impact on the environment. In the 1980s – 2000s only, about 100 million hectares of forest were cut for agricultural fields, which is equivalent to the total area of Germany and France. For nature, this is a huge loss.
That’s why environmentalists are calling for a rethink of human action and pushing for a biodiversity framework. Activists stress the necessity of building a new economy where sustainable use of natural resources and human and planetary health come first.
AI can become a tool to achieve “green” goals. The technology is used by Kew’s Millennium Seed Bank (MSBP) program to assess the risks of plant extinction. AI will signal a possible extinction in time so that researchers can take action in advance and protect a certain species.
To determine which plants need to be protected first, they create AI-based analytical programs, like Captain. The neural network analyzes plant species and evaluates the budget allocated for nature protection. Considering how people use land, it offers the most effective model for protecting biodiversity. That is, it suggests which type of plant is better to protect to get great benefits for nature.
The UK Center for Ecology & Hydrology (UKCEH) is working with Keen AI on an AI system that can scan roadside plants. Among the local flora, the program will recognize alien species that were accidentally introduced from another country and violate the ecosystem formed over the centuries. Such a measure will preserve the unique ecological system and reduce the cost of environmental research.
- Source: aiworldschool.com
AI and solving the water crisis
The WHO has estimated that around 2 billion people in the world drink unclean water, and more than 700 million people have no access to drinking water at all. According to some forecasts, by 2025, half of the population will live in regions with water scarcity.
The method of extracting salt from seawater, proposed by Martin Bazant, Professor of Engineering and Mathematics, partially solves the issue of water scarcity but at the same time pollutes the air even more. To create a shock wave that “knocks out” salt from water, too much electricity is needed. Therefore, humanity is looking for alternative “green” methods of regulating water consumption and resuming its reserves.
AI systems can become a silver bullet in the fight against drinking water scarcity. For example, researchers at the University of Chicago are developing AI-based software that will help treat wastewater for reuse. Scientists have invented a new water resource recovery system that will reduce energy consumption. Engineers plan to introduce wireless sensors to monitor water quality around the clock.
Researchers are also training AI to find chemicals and waste in the water that harm the environment. There are over 4,000 contaminants in the PFAS family alone. And the list is much longer. To purify water from harmful substances, you need a lot of money, sophisticated equipment, and specialists. Therefore, researchers want to use AI to create better chemical probes for water purification.
AI and air quality regulation
The WHO has calculated that 99% of the world’s population breathes air that exceeds all standards set by the organization for the content of harmful substances. According to the Institute for Health Metrics and Evaluation (IHME), polluted air causes about 12% of annual deaths.
- Source: ourworldindata.org
Dirty air causes heart disease, lower respiratory tract infections, lung cancer, and chronic obstructive pulmonary disease (COPD). Its influence is so strong that IHME researchers rank it third after high blood pressure and smoking.
Scientists also have high hopes for AI in helping to regulate the purity of the air. Even though there are more than 14,000 measuring stations around the world, there are enough “gaps” on the ground. To increase the coverage of the territory, engineers can use IoT devices that will send data to an AI application. The algorithm will process the data and transmit information about the pollution level in a particular area.
The smart algorithm claims to be a cheaper and more efficient way to predict air pollution levels in a region. An AI system will provide a more accurate map of the degree of air purity so that the administration can find out the reasons for the concentration of harmful substances and pinpoint this problem. AI will make information publicly available so that people with lung diseases can determine where they should buy houses, work, and walk.
An example of a working model is the system created by the University of Houston. AI predicts ozone pollution levels for up to two weeks. The model analyzes the set of factors that affect air purity and predicts what exactly will lead to pollution. For example, if AI indicates that the level of ozone in the atmosphere will increase due to traffic congestion, the administration can limit the number of cars on the road.
Conclusion
Many of the above examples of the use of AI in environmental sustainability seem utopian. And they seem more fiction than the truth. But with the right application, AI will become the “straw” that will save the “drowning.” PwC estimates that by 2030 the use of AI in environmental programs will bring more than $5 trillion into the global economy. This amount is 4.4% higher than what the average business brings. If we believe in this technology, it will change our world for the better.