The Industrial Revolution marked the beginning of a degree of technological innovation humankind had never witnessed before, but at the same time, it also set off a time bomb that is ticking, waiting to extinguish life on earth as we know it – climate change. The industrial era kickstarted an unusual increase in the human population and greenhouse gas concentrations in the atmosphere. As years passed by, the holes in the Ozone got bigger and bigger, the temperature and sea-level rose faster and faster. The loss of natural habitat resulted in the extinction of a massive number of species. Climate change is one of the significant issues talked about in every part of the world. The recent COVID-19 pandemic showed us how we are still incapable of managing civilization-ending threats like climate change.
With the rise of the powerful technology of Artificial Intelligence (AI), researchers are trying to find solutions addressing various issues related to climate change.
Artificial Intelligence has the potential to help combat climate change in a variety of ways. Although AI is not a perfect solution, it represents a step toward a better future. But to what extend can AI play a role in saving our planet? Is AI really a friend or a foe in fighting against climate change?
Infrastructure is perhaps one of the most obvious areas in which AI could make significant improvements. The global infrastructure is a vast and complex network, but its tendrils stretch down to the smallest of levels. It is dependent on a wide variety of factors, from weather to politics to the whims of individual human beings, that is impossible for humans to comprehend. For AI systems, however, comprehension and even prediction might be feasible.
Machine learning-based models could, for instance, help predict how much of which goods will be needed in a specific country, city, or even shops. The supply chain could be optimized in such a way where emissions are minimized. Goods can be dynamically bundled together to reduce the necessary number of ships or trucks, and due to accurate demand predictions, the waste is minimized. Less waste means fewer resources need to be extracted and less energy required to produce, transport, and destroy the products.
Humans are the only creatures that generate tons of small and large amounts of waste in this world. As the population is expected to increase year by year, so will be the amount of waste produced. The improper management of solid waste poses a threat to the environment and health. Using AI technologies, proper planning of Solid Waste Management(SWM) strategies can be adopted. AI can be employed to “forecast waste generation patterns, optimize waste collection truck routes, locate waste management facilities, and simulate waste conversion processes.” Using various classification models, solid waste can be segregated on arrival much more efficiently for recycling. This also protects the workers from various health hazards as they have to work in these toxic environments.
Energy infrastructure could also greatly benefit from AI techniques. Energy supply and demand, especially when considering renewable energy sources dependant on the weather conditions, are highly unpredictable. By accurately forecasting these variables, the need for polluting standby power plants and inefficient energy storage is minimised. Machine learning techniques are already being used to improve predictions of energy supply and demand, and as more relevant variables are incorporated in the models, these forecasts are set to improve. Artificial intelligence could also allow buildings to directly communicate with the grid to adapt to the fluctuating energy supply that intermittent energy sources might cause. In fact, Artificial Intelligence could take charge of the energy consumption of an entire buildings to dynamically adapt to environmental conditions and building occupancy. Google, for example, was able to drastically reduce the energy consumption of the cooling systems of its data centres by handing their control over to an AI system.
Discovering New Materials
Scientific discoveries lie at the root of many of the leaps humanity has made over the last few centuries. However, the scientific process is far from deterministic, a significant number of historical discoveries were to some extent the product of chance. This raises the question of which society-changing discoveries we have missed out as a result of an unfortunate roll of the dice. Part of the problem is that there is a near-endless number of substances and ways of combining and manipulating them, making an exhaustive search for new substances and materials impossible. Artificial intelligence could revolutionize this process: by training a neural network to recognize potentially fruitful substances. Scientists can start their research with a good idea of where to look and thus increase their chances of making a new groundbreaking discovery. This has a myriad of consequences for climate change where its application can help us invent more sustainable building materials, solar fuels, or batteries. Solar fuels would turn solar energy from an intermittent energy source into a reliable, constant energy source, and batteries could help improve electric automobiles and keep the electricity grid stable. Machine Learning might also accelerate the development of carbon-free nuclear fusion energy by suggesting experimentation parameters.
Monitoring Climate Change
In order to combat climate change, it is important that we understand it as best as we can. Artificial Intelligence can help us build better climate models, especially models that use global data to make predictions at the local level. Satellite images could be analyzed using AI to determine where air pollution and carbon emissions are coming from. This means that coal plants and other polluting industries can be directly held accountable for the pollution they generate. At the more local level, AI can help cities keep track of when and where energy is consumed through computer vision analysis of satellite imagery, thus identifying areas where there is room for improvement. The same technique could be used to monitor deforestation and detect illegal logging activity. Forests act as carbon sinks, and deforestation is responsible for up to 10% of global greenhouse gas emissions. Furthermore, forests are often cleared to make room for polluting industries such as livestock.
A perhaps surprising field in which AI could aid in is a move towards sustainability in agriculture. This vital industry is responsible for 20% of worldwide CO2 emissions and is also being threatened by climate change. The vast majority of crops that are sown in massive fields are of the same kind, which is called monoculture farming. Though this type of farming is cheap and efficient, it has severe climate impacts. It also results in soil infertility. The alternative, cultivating multiple types of crops in the same field is better for soil quality, biodiversity, and the environment at large, but it does not easily scale up. One of the main challenges in harvesting crops is that different crops can have vastly different shapes and sizes and has to be reaped at different times. So a machine/farmer has to selectively pick up ripe fruits and vegetables whilst leaving under-ripe produce in the ground. This is where robot farmers come in: a combination of robotics and computer vision technology could create machines that efficiently harvest the right crops from environmentally friendly permaculture fields at an industrial scale. Farming robots could also remove weeds mechanically and control pests in a sustainable way. A move away from monoculture farming also reduces the need for artificial fertilizers, the production of which is currently dependant on fossil fuels.
Fossil Fuel Industry
Of course access to Artificial Intelligence is not restricted to environmentally friendly companies and industries. The cloud computing industry has grown rapidly over the past few years, and the oil industry, consisting of some of the world’s wealthiest companies, is a vastly lucrative potential market. It is unsurprising then, that despite their green ambitions and backlash from employees, tech companies have been competing to strike deals with the fossil fuel industry. The fossil fuel industry now makes up about 10% of the cloud computing market and pays the tech industry about $20 billion each year in exchange for their resources and expertise. Amongst the services provided are AI and Machine Learning solutions to some of the oil industry’s biggest conundrums. For example, oil companies use seismic surveys to discover new oil fields. These surveys produce massive amounts of data that is hard to interpret but could be interpreted much more quickly and efficiently using Machine Learning. Artificial Intelligence can also help improve the oil extraction process, making it more efficient. As oil extraction becomes more efficient and the discovery of new oil fields becomes easier, the oil price will drop, hampering the renewable energy transition.
It has to be noted that the fossil fuel industry has a long history of using its financial, political, and societal influence to slow down any change that might impact their profits. If AI is effective enough in reducing global energy consumption, the industry might very well use its influence to slow down this application of Artificial Intelligence.
However, we have to keep in mind that artificial intelligence technology is not the only dark cloud on the horizon for the fossil fuel industry. Renewable energy technologies are becoming cheaper and more efficient. In fact, the price of solar energy has recently dropped below oil and gas prices in many countries. Politically, the oil and gas industry are also losing power. The Biden administration in the US is less than a month old, but it is already making good on its campaign promise to transition away from oil towards renewables. President Biden has halted the construction of the controversial Keystone XL pipeline, rejoined the Paris Agreement, suspended the sale of oil and gas leases on federal land, cut subsidies for the fossil fuel industry, and even created not one but two new climate positions in government. This comes on top of the heavy blow the Covid-19 pandemic has dealt to the industry. Some have even predicted that this decade might see the collapse of the industry. At any rate, it is clear that the fossil fuel industry is losing power, and the tight grip it has traditionally had on the energy market is slipping. It is doubtful that the increased efficiency resulting from the newfound camaraderie between Big Tech and Big Oil will be able to turn these tides of social, political, and technological change, let alone negate the myriad of ways in which AI can contribute to the fight against climate change.
Bots and Misinformation
Another way in which Artificial Intelligence could spell out bad news for the fight against climate change is the AI-enabled spread of disinformation through online platforms. Twitter bots have been found to be a major source of misinformation on climate change. In fact, a recent report revealed that, on average, Twitter bots were responsible for a quarter of all tweets about the climate crisis, the vast majority of which present a negative opinion on climate action. Social media bots allow third parties to wield a significant influence on public opinion on climate action, which, in a democracy could form a considerable obstacle for the introduction of pro-climate policies and the election of pro-climate politicians.
As AI can be used to spread misinformation, however, it can also be used to combat it. Technologies using AI have been trained to successfully distinguish between Twitter activity of humans and bots. Though this approach has its limitations, it can help curb the spread of online misinformation. Another sign of hope comes in the form of a recent survey on public opinion of climate change, the largest ever. It found that 64% of the world’s population view climate change as an emergency. So despite efforts to steer public opinion away from climate action, climate change is widely considered an emergency.
Though Artificial Intelligence can help us save energy, it also consumes a lot of energy itself. Machine learning in particular requires a lot of energy. A report from 2019 estimated that the CO2 emissions associated with training a neural network are equivalent to the entire life cycle emissions of five cars. This is an unfortunate fact for AI as a green technology, but the industry is well aware of the problem and is working on solutions to drastically bring down the energy consumption of AI. Additionally, the tech industry is one of the fastest adopters of renewable energy. With data centers increasingly running on renewable energy and Artificial Intelligence using less and less energy, the road is clear for Artificial intelligence to help us achieve our climate goals.