AI and Climate Change

Summary: Discussing some interesting ways AI technology is being applied to help people measure, predict, and target climate change.

Introduction

Artificial intelligence has been revolutionizing industries across the globe over the last couple years, causing all kinds of changes to how we interact with and study the world. One major topic on people’s minds is how AI will impact climate change in the near future, and whether it’s going to do more good than bad. Today, we’re going to talk about some specific topics where AI is playing a large role in the climate and what some benefits and risks might be.

First, let’s talk about what the current state of climate change is. If the amount of greenhouse gas emissions remain at the levels they are today, by 2100, average global temperatures could increase by 4° C (7.2° F). This will cause a cascade of problems including the oceans becoming increasingly acidic, water levels rising, more frequent and more severe weather, and other changes that will have a huge impact on the entire planet.

While this sounds bleak, there are things we can do to reduce the scale of climate change. Current models of how global temperatures are changing strongly depend on emissions of greenhouse gases. If we can stop increasing the amount of carbon dioxide released into the atmosphere by 2050, global average temperatures may only increase by about 1° C (1.8 ° F) from now until the end of the century. This would still have a huge impact on the climate, but the changes would be less catastrophic and happen slower, people would have more time to prepare and adapt, and more technology could be developed to help protect the planet – technology like AI.

AI tech is already playing a major role in the changing climate, which we’ll talk about in more detail next. We’re going to talk about how AI is currently being used to study how the planet’s climate is changing, target climate interventions, and what ethical complications are arising with these new applications.

Collecting data

One of the most important steps in responding to the current climate crisis is collecting the right data from the right places at the right time. This includes measurements of ocean temperatures across the globe and at varying depths, mapping ocean currents, studying cloud patterns, sampling ice cores, measuring gases in the atmosphere, and so much more. All of this data combined is used to develop and validate models that help scientists predict weather and climate at local and global scales.

So where does AI show up in data collection? In a lot of places, as it turns out, including the bottom of the ocean. Collecting temperature measurements at different depths in the ocean is really important for studying how the ocean is absorbing carbon dioxide, a greenhouse gas, and how heat is being transported throughout it. Both of these are critical for scientists to have up-to-date information on, as changes to both can have a drastic, domino effect on ocean ecosystems and the atmosphere.

The main problem is that reaching and navigating the ocean floor is a complicated and dangerous task, involving equipment that can handle extreme pressures and temperatures, navigate without relying on GPS or visual cues, and more. It’s expensive, time-consuming, and difficult to get even a little amount of data about the deep ocean. However, AI could be used to control bots that autonomously collect that data over long stretches of time without human input. It could open up a lot of doors for improving climate models and deepen our understanding of the world’s oceans.

Another place that’s hard for people to reach is Antarctica, and it’s also a place that holds critical information about the climate. For about six months of the year, it’s too cold and dangerous for ships to reach Antarctica due to the ice in the ocean, which means no ground measurements can be taken. Having ground measurements throughout the entire year would help with studying melting ice and other ways the climate is changing in a very important location, and like the seafloor, AI could be used to fill this gap by autonomously collecting measurements in places humans can’t go safely.

Analyzing climate data

There are other ways AI can help with studying the climate, including improving data analysis of existing climate data. Currently, satellites, ground measurements, and other sources of data provide more information than there are people to analyze and process it, even with the help of supercomputers. AI can step in here to examine large amounts of data and find new trends for scientists to look into. This can help develop better climate models that are more accurate and detailed.

Speeding up data analysis and finding new patterns can have impacts beyond better weather and climate predictions. One example is Project Sunroof by Google, an AI tool that analyzes data to estimate how much a homeowner could save on their energy bill by installing solar panels, customized by their location. This can help incentivize people to choose greener options for their energy needs.

Another example of how AI is analyzing data to help combat climate change is IBM’s Green Horizons initiative, which partners with organizations in China to pinpoint sources of air pollution down to individual streets. This information can help alert people to air quality issues and address specific sources of air pollution. It’s more efficient and effective to prioritize smaller interventions that can have a large impact instead of implementing broad, blanket measures.

One other way AI is becoming intertwined with climate management is through studying how crops and trees are dispersed. This information is important because it can improve farming efficiency, and so understanding what is planted where can help people make more informed decisions about what to grow.

AI is also being used to study deforestation through Microsoft AI for Earth, an initiative to find more applications for AI that combat climate change. Having up-to-date information about deforestation means people can focus conservation efforts and planting trees in the locations that need it the most.

Affecting the climate

So far, we’ve talked about how AI is helping to collect and analyze all kinds of data studying the planet, but what about taking action? Taking the data AI is providing and transforming it into targeted interventions can work at both local and global scales to improve efficiency, reduce emissions, and pursue greener alternatives.

One example of how AI can cut down on electricity waste is through power grid optimization. When determining how much power to generate, people have to overestimate to ensure there’s enough power to meet a community’s needs. However, generating extra power usually means some amount of waste or extra emissions released into the atmosphere. AI can analyze a community’s power usage combined with data about local power generation to create estimates that are more finely tuned than current ones. Using this information, power plants can cut down on making electricity that isn’t used.

AI can also drastically advance reforestation efforts through the use of drones combined with data analysis of remote areas that are difficult for humans to reach. Sending drones to plant trees in places where it’s expensive, dangerous, or just difficult for people to go opens up a lot more opportunities to plant trees that are likely to be undisturbed by deforestation.

There are a lot of ways AI can create improvements to improve efficiency in industries like aerospace, construction, and manufacturing. AI’s ability to extrapolate patterns from large datasets can be applied to engineering designs to find improvements that can reduce weight, increase strength, and make designs more resilient and efficient. This has wide-reaching implications, including reducing emissions for aircraft and other transportation methods.

Preventative maintenance is yet another instance of harnessing AI to reduce wasted energy. Heating and cooling systems require a lot of energy, as anyone paying an electric bill after blasting the air conditioning all summer knows. Food and medicine temperature control systems also rely on a lot of electricity. AI can study when these systems need to be running and what levels they need to run at, and automatically control them so they’re only using the energy they need.

Complications

We’ve discussed a lot of examples of how AI is having a positive impact on climate change efforts, but there are some ways it can be used to make matters worse – including AI’s carbon footprint and how AI handles the potential for catastrophic failures.

One of the biggest concerns with the recent AI tech boom is the amount of energy required by AI systems to run. A lot of AI technology requires some seriously energy-intense computations, which in turn have an increasing carbon footprint. Currently, data centers are taking up 1.3% of global energy use and increases in AI energy requirements have driven that amount higher and higher for the last couple years, and as more large corporations dive into AI tech, that number is only going to rise.

Managing the carbon footprint from AI is currently limited by large corporations. For instance, Google is contributing about 10% of the total amount of energy cost for data centers across the globe. According to the company’s 2024 Environmental Report, their total greenhouse gas emissions increased by 13% in one year mostly due to their increased use of AI. Furthermore, the report states that Google may find it challenging to reduce emissions going forward due to AI energy requirements. To put this in more context, Google uses 0.1% of the energy used across the entire planet, and some of the language in the environmental report they released implies they are prioritizing increasing AI computational costs over reducing emissions.

One other way AI can cause some problems with responding to the climate crisis is through how it handles situations that could be catastrophic if the wrong decision is made. AI makes decisions based on averages from large amounts of data, so most of the time it will have an acceptable answer. But some situations have outliers where the average answer would be wrong. An example could be calculating power requirements for a grid. Based off averages from large datasets, most of the time AI will come up with a reasonable answer, but if it doesn’t, a community might end up with blackouts. Figuring out how to integrate AI into monitoring systems that have the potential for critical failures is going to be a complicated problem to solve.

Conclusion

We’ve covered some examples of how AI can be used to have some big impacts on climate change, both positive and negative. There are a lot more questions to be answered moving forward – what other ways can AI be used to target climate change? How else can it be used to negatively affect the climate? Is continuing to integrate AI tech like in these examples worth it? What else can be done to change how our future will look? Ask your friends!