Click to learn more about author Pranay Agrawal.

In a landmark move, earlier this year the Biden administration and other global leaders laid out ambitious pledges aimed at reducing emissions over the coming decades. The U.S. announced that it would reduce greenhouse gas emissions by 50% by the end of the decade, while other countries also announced aggressive climate targets. And although these revised goals and targets have resulted in a lot of goodwill and positivity, it is easy to be overwhelmed when it comes to figuring out how exactly we are going to meet these goals in such a short time. Thus, governments and corporations are likely to lean more heavily on technology to speed up innovation.

Naturally, emerging technologies – particularly AI – have been touted as one of the primary tools in helping to drive rapid emissions reduction efforts. However, despite the appeal of fields like deep learning and automation, it would be misguided and inefficient to use AI resources in a race to build supposedly “better” solutions – such as mechanized and synthetic carbon capture – instead of using them strategically to correct existing and long-overlooked issues. As we look to the future, we need to ensure that the cost of using AI to tackle climate change does not overshadow its benefits.

There are many areas, such as agriculture, energy, and transportation, that need serious emissions help. But to truly wield all the potential AI has to fight climate change, we also have to think about the way we deploy AI and the ethics behind it – not just the problems we need to solve.

With that in mind, here are several areas and used cases to demonstrate the use of #AI4Good and what we must do to rethink the way we approach using AI moving forward, so that we are able to develop a holistic climate-environment partnership.

Rethinking AI Ethics

To really use AI as well as possible, AI technology providers and their clients need to not just think about the issues we are trying to solve with AI but the way we use AI itself. For example, it is second nature for AI providers to ask potential clients if they plan to use their technology to perpetrate a crime or use the technology for other nefarious means. Why aren’t we doing the same when it comes to how customers plan on using technology in relation to the environment? This needs to become one of the first questions we ask clients moving forward, and if there are negative consequences on environmental health, then a project should not proceed.

Furthermore, AI companies shouldn’t solely rely on client usage as the determining factor in whether a project proceeds or not. Instead, they themselves should prioritize projects that do not have adverse benefits on the environment by adopting an AI code of climate ethics. Moreover, this should not be a bit of lip service – AI climate ethics needs to be formally ingrained and adhering to these ethics needs to be binding if we are truly going to use AI for climate good. Once that occurs, we can begin to tackle the real-world issues that we face, including:

1. Using AI for sustainable agriculture: Sustainable agriculture is going to need to be a huge component of the world’s climate action plans. But plants make up only half of the equation; the other half is the soil itself. Next to the ocean, soil is the world’s biggest carbon sink. In fact, according to the Food and Agriculture Organization of the United Nations (FAO), the top 30 centimeters of topsoil contains nearly double the amount of carbon that can be found in all of Earth’s atmosphere. Yet over-farming, poor soil maintenance, and poor planting practices – along with the impact climate change itself is having on soil – continue to undercut the potential carbon-capture productivity of innumerable acres of farmland on a yearly basis. Additionally, much of modern-day agriculture relies heavily on the use of nitrogen-based fertilizers, which can convert into nitrous oxide, a greenhouse gas 300 times more potent than carbon dioxide. 

The solution: AI technology can synthesize real-time data during planting season to ensure that crops are positioned in the best possible location so that unneeded wear and tear on soil can be avoided. Furthermore, AI can be used to study factors such as microbial health and diagnose issues so that immediate steps can be taken to rectify the situation during both planting and fallow seasons. Robots that run on machine learning software could help farmers manage a mix of crops more effectively at scale, while algorithms could help farmers predict what crops to plant when, regenerating the health of their land and reducing the need for fertilizers. 

2. Using AI for better energy management: AI can be used to monitor how efficiently buildings are using energy and identify urban heat islands – areas where urban construction leads to more heating than its surroundings, leading to more electricity consumption to cool it down. Identification of such areas can serve as an input for encouraging more efficient and environment-conscious planning. If we’re going to rely on more renewable energy sources, utilities will need better ways of predicting how much energy is needed, in real time and over the long term. Algorithms already exist that can forecast energy demand, but they could be improved by taking into account finer local weather and climate patterns or household behavior. Efforts to make the algorithms more explainable could also help utility operators interpret their outputs and use them in scheduling when to bring renewable sources online.

3. Using AI for efficient transport systems: All three transport systems – land, sea, and air – are prime contributors to carbon emissions. Talking about land transportation first: Traffic congestion leading to engine idling, use of vehicles beyond a permissible lifetime, and inefficient rail systems increasing reliance on electricity are all leading causes of high emissions. Speaking of sea transportation, shipping emissions account for 2.5% of global greenhouse gas (GHG) emissions and could rise by as much as 250% by 2050 under a business-as-usual scenario, according to the International Maritime Organization (IMO). Finally, commercial aviation contributes to approximately 5% of the world’s climate-warming problem and by 2050, commercial aircraft emissions could triple given the projected growth of passenger air travel and freight, according to Environment and Energy Study Institute.

This is obviously a huge challenge to tackle, but AI can factor in to drive immediate results and get momentum building. For example, AI and machine learning can be employed to improve vehicle flow. In Pittsburgh, AI has been used to reduce engine idling by 40%, delivering an estimated 20% cut in carbon emissions. AI algorithms can be deployed for traffic light sequencing according to real-time information on traffic volumes, collected from cameras and radar. In a similar way, collection of weather data will enable ship crew to plot the optimum speed to arrive in port when scheduled, saving thousands of tons of fuel across a trip and also aid in suggesting the shortest route. 

4. Tackling AI and Computing’s Footprint: If governments and technology companies are going to lean upon AI and computing power as heavily as they will likely need to over the next several decades to fight climate change, they need to address the elephant in the room: technology’s carbon footprint. With more devices coming online every day, and more intensive computing being done – including everything from gaming to bitcoin “mining” – technology itself is having major ramifications on global climate health. In fact, a recent study estimates that 1% of the world’s electricity usage was generated by data centers alone in 2018. And although these levels have remained relatively stable thanks to energy-saving innovations – see Deepmind’s work in reducing Google’s footprint – as AI is pushed to tackle more sophisticated tasks, energy usage will likely ramp up significantly as a result. So, what can be done?

In the immediate term, companies need to double down on their investments in areas such as cloud and quantum computing – which have lower electricity demands than traditional data center computing – and begin mandating that their energy needs are met solely by renewable energy. However, in the long term, AI companies need to rethink the way AI functions by studying outside areas – in particular, neuroscience.

The human brain is one of the most powerful tools on the planet and uses just a fraction of the energy that a data center does. Also, areas like neuroscience are underleveraged to solve problems holistically and effectively, which helps break the technology silos. 

AI is already one of the most exciting technology sectors today. And as the world makes a huge effort towards combating climate change, it stands to be one of the biggest allies in helping the world address both the climate challenges of today as well as those of tomorrow.