Is AI Humanity's Doom or Saviour?

By Gemma Tabet

Written: June 12th 2023

Theme of Issue: SDG 13, Climate Action. Here is the official UN link where you can learn more about this particular Sustainable Development Goal: https://sdgs.un.org/goals/goal13

---

Photo by Markus Spiske via Unsplash

---

With recent technological developments, it is not unlikely that in today’s modern era, the word ‘AI’ or artificial intelligence, evokes the dystopian idea of a grim, robot-ruled world with mind-controlled human slaves. Yet, although AI could potentially become a threat if not used safely and responsibly, it can also be a tool for good. Sustainable Development Goal 13, Climate Action, might be achieved with the help of AI. Climate change, unlike AI, is an already existing threat for many, primarily created by global warming, rapid urbanization, and industrial revolution, causing, in turn, the destruction of flora, fauna, and human beings. While there are possible solutions to alleviate the situation, these might become weaker and ineffective over time as the severity increases, making the need to find advanced ways of predicting, analyzing, monitoring and mitigating climate change only more urgent. And to do this, AI might be the only answer.

The term ‘artificial intelligence’ was first coined in 1956, during a conference at Dartmouth College in Hanover, New Hampshire, and although the years after were plagued by periods of ‘AI winter’ when progress in machine learning slowed, by 2023, it is evident to many that the field is now deeply integrated in almost every field. AI can calculate, use data and learn algorithms in a way that is impossible for even the brightest humans, making it ideal for the predictions and decision making needed to mitigate the impacts of climate change. It’s able to make these predictions using trends, patterns, and extensive data sets, consistently keeping up with ever evolving changes in the environment to create detailed predictions and models that help policymakers take effective decisions and deploy earlier mitigation efforts. Some of AI’s potential and current uses in the fight against climate change can be seen below:

1. Green Transportation

Any form of human transport, whether a train, car or plane, requires the use of fossil fuels, and makes up one fifth of global CO2 emissions. However, AI-driven vehicles could reduce this number by minimizing energy when choosing the most efficient routes. Some studies predict AI self-driving cars could reduce greenhouse gas emissions by 50% in 2050.

2. Sustainable Cities

AI could make cities smarter and more sustainable, by providing information and predictions on where, when and how a climate disaster could impact an area in the future. This can help companies such as 'Sipremo' find concrete solutions to mitigate the effects or even avoid the disaster, by improving cities to better fit the evolving environment. After a disaster, AI could make disaster response more efficient by processing information faster and directing resources to the most vulnerable areas. AI cities could also optimize water and energy uses, a tool already used in some cities in Brazil and the Philippines.

3. Smart Agriculture

Farming alone accounts for 70% of global freshwater use, and due to poorly managed systems, around 40% of this is lost. AI could help prevent this, by monitoring agricultural inputs and outputs to irrigate crops more efficiently. AI can optimize fertilizer use and schedule better planting seasons which would reduce greenhouse emissions and lead to more productive harvests with better yield. For example, in India, peanut farmers using AI have achieved 30% higher yields.

4. Fighting Plastic Pollution

New technologies that use AI are helping to create detailed maps of plastic densities in the ocean, detecting and monitoring plastic debris more accurately. This helps companies such as the 'Ocean Cleanup' focus their cleaning efforts on the most impacted areas, or the plastic ‘hot spots’.

5. Better Conservation

AI is getting better at processing visual information and because it can analyze satellite images faster than humans, it can provide conclusions to help conservation efforts, by identifying if certain areas of a coral reef are dying or if certain areas of a forest are under threat of deforestation. Not only, but it can also offer predictions of the potential of deforestation using information such as distance to water sources or cities.

6. Weather Forecasting

With AI, meteorologists and scientists are able to predict weather changes with 99% accuracy. These better predictions would allow cities and people to be more prepared for storms and other natural disasters. This technology, which uses deep learning to find new patterns in data, can help with the prediction of wildfires, heat waves, and more. For example, the start-up 'Kettle' was able to develop an AI fire model that could accurately predict wildfires in the top 20% of at-risk areas. Yet, despite AI having countless potential uses to help fight climate change, it also brings a host of potential negative effects that could worsen global warming and harm the environment more than help it, seen by key issues in two main parts of the technology’s life cycle.

The first element of this cycle is manufacturing. AI technologies often require rare resources such as tantalum, lithium and cobalt, whose extraction presents threats to both humans and the environment. The mining waste from tantalum, for example, has polluted local water resources in the Democratic Republic of Congo. Yet it is lithium, found in rechargeable batteries, that provides an even greater issue. It requires the use of 500, 000 gallons of water for every ton of lithium extracted, which is roughly the same as filling an Olympic-sized swimming pool. In places like Chile, which is the second largest producer of lithium in the world, indigenous communities such as the Copiapó people clash with mining companies over the water loss damages caused by lithium extraction, which affects both flora and fauna species, as well as local populations. Lithium mining in Salar de Atacama, for example, consumes 65% of the region’s water, according to the Institute for Energy Research, proving that further progress in the AI revolution will evidently make already marginalized communities suffer more.

The second element of the cycle is usage. AI consumes huge amounts of energy, emitting emissions similar to those of the aviation industry. But another problem lies with training the AI model, a great time and energy consuming process. Training an AI model requires billions of training examples and countless training cycles, and a paper by the University of Massachusetts Amherst found that “training a single AI model can emit as much carbon as five cars in their lifetimes” for only one training run. This is similar to the carbon emissions emitted by fifty-six people in a year, but these numbers are inaccurate, as most AI models require multiple training runs to improve and develop, having a much greater energy usage. Training GPT-3, for example, which is a highly advanced AI language program and the ‘mother’ of ChatGPT, took 1.287 gigawatt hours, which is the same as the electricity consumed by 120 U.S. homes in year, generating 502 tons of carbon emissions, the same as 110 U.S. cars in a year. Moreover, most data centers use GPUs or graphic processing units to train AI models, which are among the most energy demanding chips in the world. In fact, some scientists are even beginning to believe that future studies will reveal GPUs burn as much energy and power as a small country.

In conclusion, AI has the potential to act as one of the world’s most powerful technologies in fighting climate change. Already today it is being used in countless fields, from satellite imagery to more efficient agriculture, and is revealing to be an essential tool that can unlock key solutions to better mitigate and survive an ever-changing environment. Policymakers, companies, and governments must now seriously consider and evaluate solutions to ensure that a further use of AI will not revert its role as an key tool in the fight of climate change by turning it into a main contributor of greenhouse gas emissions and water scarcity. Priority should be given to AI research that is less energy-consuming, such as automated reasoning, as well as having companies provide consistent reports on the costs of AI training for transparency. As Andrea Renda, member of an expert group advising the European Commission, clearly states, it is time to merge these two debates, “One is on digital technology and the other one is on sustainable development, and in particular the environment. If we use the former to save the latter, I think we will have made the best possible use of the resources that we have... Otherwise, we're just wasting time.”