AI’s Environmental Impacts

” Pour ce qui est de l’avenir, il ne s’agit pas de le prévoir, mais de le rendre possible. “? – Antoine de Saint Exupéry, Citadelle, 1948. Translation: "it is important not merely to foresee the future, but to make it possible."

Global climate change is one of the most pressing issues of our time, primarily caused by greenhouse gas emissions (more energy enters the Earth's system than is emitted). The result is a rapid warming of the atmosphere, which affects the weather significantly.

According to IPCC Special Report on Global Warming:

Human activities are estimated to have caused approximately 1.0°C of global warming?above pre-industrial levels, with a likely range of 0.8°C to 1.2°C. Global warming is likely to reach 1.5°C between 2030 and 2052 if it continues to increase at the current rate.

In terms of climate change, AI has several negative effects for the environment. These issues may not be immediately apparent to many people, but they are real and should be considered.

According to Elizabeth Kolbert of the New Yorker "Artificial intelligence requires a lot of power [...]. The kind of machine learning that produced ChatGPT relies on models that process fantastic amounts of information, and every bit of processing takes energy. When ChatGPT spits out information (or writes someone’s high-school essay), that, too, requires a lot of processing. It’s been estimated that ChatGPT is responding to something like two hundred million requests per day, and, in so doing, is consuming more than half a million kilowatt-hours of electricity. (For comparison’s sake, the average U.S. household consumes twenty-nine kilowatt-hours a day.)"

The cost of training and developing AI models is quite high. Financially, it involves the cost of hardware, electricity, and cloud compute time. Environmentally, it has a significant carbon footprint due to the energy needed to power modern tensor processing units.

But, data center construction continues to accelerate. According to Bloomberg "Microsoft said capital expenditures hit $19 billion in the quarter and will increase in the next fiscal year. Meta tweaked its capex to $37 billion to $40 billion for the full year, raising the low end of an earlier range by $2 billion. And Amazon, the market leader in cloud computing, said it spent $30.5 billion in the first half of the year and pledged to exceed that figure over the next six months. Much of that money is going to data centers, the unglamorous warehouse-like buildings filled with rows and rows of powerful computer servers, routers, cabling and cooling systems."

A study conducted by the University of Massachusetts, Amherst, found that training a single large AI model can emit over 626,000 pounds of carbon dioxide. This is equivalent to five times the lifetime emissions of an average American car, representing a substantial environmental impact. The scary part is that these numbers represent only the baselines.

MIT Technology Review reported that in practice, AI researchers are more likely to either build a new model from scratch or tweak an existing one to fit a new dataset. Both of these options usually mean a lot more rounds of training and fine-tuning will be required. The computation cost and environmental impact can go up even more if the model keeps learning continuously. Again, deep learning methods typically require substantial amounts of data. This data must be acquired, transferred, stored, and processed, all of which necessitate equipment and energy, thereby impacting the environment.

International Energy Agency reported in 2023 that global CO2 emissions reached more than 37 billion metric tons, with data centers and transmission networks accounting for 1-1.5% of global electricity consumption. This number could skyrocket quickly with the ongoing rapid growth of AI.

According to Harvard Business Review "all these environmental impacts are expected to escalate considerably, with the global AI energy demand projected to exponentially increase to at least 10 times the current level and exceed the annual electricity consumption of a small country like Belgium by 2026. In the United States, the rapidly growing AI demand is poised to drive data center energy consumption to about 6% of the nation’s total electricity usage in 2026, adding further pressure on grid infrastructures and highlighting the urgent need for sustainable solutions to support continued AI advancement."

AI's Positive Impact on the Environment

The use of A.I. could potentially help eliminate some of the problems it creates. It could speed up tech advancements, help develop new sustainable materials, boost energy efficiency, aid in disaster response.

For example, AI's forecasting abilities may be useful for flood management, storm responses, and reducing global carbon emissions. AI can predict environmental processes like erosion, landslides, and weather patterns, and offer strategies for electricity optimization. Using A.I., complex forecasts can be generated in real-time to model future scenarios, help with climate resilience and adaptation, and be used to track progress on pollution prevention or waste reduction initiatives, such as air quality and corporate carbon footprint reductions.

Other examples of where AI can be useful to help the environment include self driving cars, which run on AI, help cut down on emissions. AI has been a game-changer in agriculture by boosting crop yields. Plus, AI can analyze satellite images to spot disaster-hit areas, making sure help gets there fast.

When it comes to enforcing environmental regulations, artificial intelligence can be exceptionally useful at forecasting which facilities will violate them (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3613804). A?study showed that in comparison to just a random selection of facilities, the US Environmental Protection Agency (EPA) could improve its detection of water pollution violators by over 600% by using machine learning algorithms.

What is Next?

AI can be used as a tool to mitigate some of the environmental damage it causes, but it will certainly not be enough to compensate for the damage it causes in order to meet the growing electricity needs of the technology. Sam Altman, the head of OpenAI said "I think we still don’t appreciate the energy needs of this technology." He further explained that he didn’t see how those needs could be met "without a breakthrough. We need fusion or radically cheaper solar plus storage, or something, at a massive scale — a scale that no one is really planning for." And that, of course, is the problem. How fast can these breakthroughs be achieved, if at all?

“Optimizing actions for a restricted set of parameters (profit, job security, etc.) without consideration of the[...] wider impacts can lead to consequences for others, including one’s future self as well as future generations.” (Walsh, T.; Evatt, A.; de Witt, C.S. Artificial Intelligence & Climate Change: Supplementary Impact Report; Technical Report; University of Oxford: Oxford, UK, 2020).

The status quo is not sustainable and something needs to be done if we want our children to have a future. As a solution to A.I.'s negative environmental impacts, many organizations are exploring alternative energy sources, such as batteries, to power their systems. But, some of these alternatives come with their own environmental challenges. For instance, lithium batteries require a ton of water for extraction, which can negatively impact the environment.

But, there are efforts underway to reduce this footprint, including the CODES Action Plan for a Sustainable Planet in the Digital Age- an initiative based on the UN Secretary General's Roadmap for Digital Cooperation.

It's not all doom and gloom, however. Although it is forecasted that AI's electricity requirements will most likely double in the next coming years, the technology will become more energy efficient. As Steve Lohr of New York Times wrote yesterday, "There is a historical precedent. Consider what happened with cloud computing. There was a surge in energy consumption in the early 2000s. And there were concerns that the increase would continue. But while the computing output of the world’s data centers jumped sixfold from 2010 to 2018, energy consumption rose only 6 percent."

The technology giants are heavily investing in alternative energy sources. They are moving their their servers to the Nordics centers to reduce their data centers' electricity consumption.

If we succeed in discovering alternative energy sources and making AI more energy-efficient fast enough, AI might just turn out to be the perfect solution for the environment after all.

Let's wait and see.

Looking Ahead

In the next issue, we will discuss AI and IP Law.

Thank you for joining me on this exploration of AI and law. Stay tuned for more in-depth analyses and discussions in my upcoming newsletters. Let's navigate this exciting and challenging landscape together.

Connect with me

I welcome your thoughts and feedback on this newsletter. Connect with me on LinkedIn to continue the conversation and stay updated on the latest developments in AI and law.

Disclaimer

The views and opinions expressed in this newsletter are solely my own and do not reflect the official policy or position of my employer, Cognizant Technology Solutions. This newsletter is an independent publication and has no affiliation with #Cognizant.