With Great Power Comes Great Responsibility
The phrase “With great power comes great responsibility” has been Spider-Man’s guiding mantra ever since he first swung into action. And just as Peter Parker navigates the fine line between his superhero persona and everyday life, this age-old principle resonates far beyond the world of Marvel – especially when it comes to Artificial Intelligence (AI). While AI doesn’t swing between NYC skyscrapers, its reach is just as expansive, weaving itself into the lives of people, their pockets, and ultimately the planet. But, for all its revolutionary potential, AI’s environmental footprint requires careful attention. Like any power this potent, it’s not just about what it can do – it’s about how we choose to wield it.
The Invisible Web
AI is no longer a futuristic idea; it's become our daily sidekick. From curating our Spotify playlists to making life-or-death decisions in healthcare, AI’s presence in our modern lives is making it all the more difficult to imagine a life without it. Yet, the infrastructure supporting AI remains hidden well behind the screens of users, until its environmental impact demands our attention.
Take AI training, for instance. A single language learning model (LLM) like our friendly neighbourhood GPT-3 can consume 1,300 megawatt hours (MWh) of electricity. For some perspective, that’s roughly equivalent to the annual energy consumption of 120 American households. In a new paper, researchers at the University of Massachusetts, Amherst, conducted a life cycle assessment of training several common large AI models. Their findings revealed that the process can generate over 626,000 pounds of carbon dioxide equivalent – nearly five times the total lifetime emissions of an average American car, including its manufacturing.
Those numbers are hard to grasp, but the impact is clear: as AI becomes more sophisticated, the energy demands multiply, leading to a growing carbon footprint that already accounts for an ever-climbing 3-4% of global emissions. With cloud computing already out-emitting commercial aviation, the increased climate burden from artificial intelligence is emerging as a main antagonist.
Water usage is another aspect where every drop counts. Cooling the data centers that run AI models like GPT-3 takes an immense 185,000 gallons of water. That’s enough to manufacture over 250 Tesla vehicles. As AI advances and the number of models increases, the demand for both energy and water will only grow. It’s like a powerful superhero whose abilities come with an unintended, and sometimes invisible cost.
An Unsung Superhero
But much like Spiderman’s potential to be more than just a costumed vigilante, AI is uniquely positioned to take on the role of a hero in the fight against climate change. Agriculture is a prime example. AI is already transforming how we manage water in farming, where overuse is a pressing issue in drought-stricken regions. With AI-powered irrigation systems, water usage can be optimized, with the potential to reduce global agricultural water consumption by as much as 25%. Smart irrigation systems are an advanced technology that helps growers monitor and optimize yield, which could be the difference between a thriving crop and a wasted harvest.
When it comes to renewable energy, AI and analytics are helping us crack the code. Solar and wind energy – while incredibly promising – have their limitations. They’re reliant on the sun shining or the wind blowing, and without proper forecasting, they can result in wasted energy or grid instability. Here’s where AI comes in: by predicting fluctuations in energy production and managing energy storage, AI helps make renewables more reliable. These tools allow for smarter, more efficient use of renewable energy, reducing reliance on fossil fuels and cutting emissions.
AI is also making a big impact in the field of climate science. With real-time data analysis, AI models can provide hyper-local climate predictions, allowing cities and countries to act quickly when natural disasters strike. Whether it's predicting hurricanes, monitoring deforestation, or tracking air pollution, AI is giving us the foresight to protect communities and ecosystems before disaster hits. In this case, it’s not just the hero we need, it’s the hero we deserve.
The Symbiote Effect
Yet, like any superhero, AI isn’t without its kryptonite. The very hardware that powers AI – servers, semiconductors, and batteries – also contributes to one of the fastest-growing environmental problems: e-waste. With typical lifespans of two to five years, computing devices demand constant hardware upgrades to keep up with their growing capabilities, leading to a mountain of discarded tech.
Depending on the adoption rate of generative AI, the technology could add 1.2 to 5 million metric tons of e-waste in total by 2030, piling on the global total of over 60 million metric tons of e-waste each year. The lithium-ion batteries and rare earth metals that power AI systems are mined in ways that not only devastate ecosystems through deforestation, water contamination, and land
degradation, but also deeply harm the communities extracting them. In regions like Zimbabwe, Namibia, and the Democratic Republic of Congo, the lithium boom has led to exploitative labour practices, forced evictions, and corruption, with local miners risking their lives for a resource that fuels global industries but leaves them impoverished. When these components are eventually discarded, the burden meets a new means. They release toxic chemicals into the ground, leaching into soil and water systems, perpetuating a vicious cycle of environmental harm.
There is always hope, though. As with Peter Parker finding innovative ways to protect the city, AI is also being leveraged to address the e-waste problem. Researchers are developing biodegradable circuit boards and sustainable battery alternatives that could reduce the environmental toll of hardware production and disposal. Google DeepMind, for example, has applied AI to improve semiconductor design, making chips more energy-efficient. In short, AI isn’t just helping us solve the world’s problems – it’s also learning how to slowly erode its own carbon footprint.
Sustainability: A New Spidey-Sense
AI’s potential is undeniable, but like any tool of great power, it requires a level of responsibility. As much as we marvel at its capabilities, we can’t afford to overlook its environmental cost. It’s up to developers, policymakers, and consumers alike to guide AI toward a sustainable future.
In recent years, major companies have begun stepping up to the plate. Microsoft, for example, has pledged to become carbon-negative by 2030. This goal was complemented by plans to implement water-free cooling systems across its new data centre developments from August 2024. The new design is projected to reduce water consumption by 125 million litres annually per data centre facility. There’s also exciting progress in areas like quantum computing and neuromorphic computing, which are new models of computing that promise to slash energy usage while boosting processing power. These innovations are the key to ensuring that AI evolves in ways that do not come at the expense of our planet.
Clearly, AI needs a framework for sustainability that promotes responsible energy use, longer-lasting hardware, and better recycling. Governments and the private sector must collaborate to ensure that AI’s future isn’t just bright – it’s green, too.