It is the world of tomorrow, beginning to be written, and at a giant pace, from today. Artificial Intelligence (AI) has already disrupted, more or less visibly, established formulas for developing all kinds of tasks. Beyond its advantages, AI unfolds a multitude of dangers that transcend ethical discussions; more and more, the way its activity impacts the environment is revealed as a threat to what it is determined to revolutionize: the world itself.
Based on the combination of algorithms, AI represents in the digital age the greatest social and economic revolution in recent centuries. The results of its operations have transformed the way many sectors operate, but more and more studies focus on the amount of energy and water it consumes, the pollution produced by the manufacturing of the components, and the electronic waste generated at the end of its useful life.
An investigation carried out by the universities of Colorado Riverside and Texas Arlington emphasizes, for example, the amount of fresh water that companies need to cool the servers dedicated to powering their systems. According to this analysis, in 2021 Google consumed 12.7 billion liters of fresh water in cooling processes. And Microsoft has acknowledged that its water consumption increased by 34% in the period 2021-2022.
“The water footprint is directly related to energy consumption; the higher consumption, the more heat generated, and the greater the cooling needs,” explains José Andrés López de Fez, a member of the Digital Society Committee of the Institute of Engineers of Spain (IIE), who adds: “The electronics at the heart of servers perform better the lower the room temperature they are in.” Thus, the United Nations Environmental Report estimates that water needs derived from AI could reach 6.6 billion cubic meters by 2027. On the other hand, it is important to note that the manufacturing of the technology that makes up the equipment involves intensive extraction of rare materials, such as silicon, germanium, gallium, boron, and phosphorus, an activity that has a significant impact on the environment and contributes to water pollution.
It is also expected that the energy consumption of data centers will multiply in the coming years. According to the International Energy Agency (IEA), 2% of global electricity demand already corresponds to these facilities. In total, the infrastructure related to AI and cryptocurrencies required 460 terawatt hours (TWh) in 2022, and projections suggest it could rise to 1,050 TWh in two years.
Europe, a pioneer in regulating AI
Few rules have managed to gather more favorable votes. The European Parliament, with 523 in favor and 46 against, has just made history by ratifying the world’s first Artificial Intelligence Act, ready to enter into force in 2026. “We are regulating as little as possible but everything necessary,” says Thierry Breton, Commissioner for the Internal Market.
The law establishes permits and prohibitions on the use of this technology based on the risk it poses to citizens, banning it in the scope of massive surveillance in public spaces but endorsing it in security actions, counter-terrorism, and prosecution of criminals. It also includes that AI systems used in the Union must be environmentally friendly and always supervised by people rather than automated systems. The European Commission is also developing other initiatives related to the ecological footprint. Thus, it plans to establish minimum efficiency requirements for servers and computers and is preparing new monitoring standards to collect and publish information on the energy and water consumption of data centers. It is also working on designing measures to promote the location of new data centers in places where residual heat can be reused to reduce energy and water needs for cooling.
Similarly, although non-binding, the United Nations General Assembly has unanimously approved a resolution for the international regulation of AI, in order to “safely and reliably harness its opportunities for sustainable development.” The United States, a promoter of this standard, argues that “private companies must be responsible” and are obligated to “close the digital divides so that everyone can access the benefits of AI.”
However, it is precisely the United States, the birthplace of the main companies developing this technology, that is dragging its feet the most on the path to implementing regulations to govern it. Although the White House has issued an executive order requiring companies to report preliminary results of their innovations to the federal Government, there is still no specific law in place, despite demands from the sector.
“It is important that we have an arbiter,” confirmed Tesla CEO Elon Musk upon leaving a meeting in September 2023, where the top names in the American technology sector gathered at the Capitol in Washington to urge the country’s politicians to act as regulators to prevent uncontrolled development of AI, “a double-edged sword” for Musk.
AI as a tool for sustainability
The IEE engineer López de Fez recalls that the first computer, the 1946 ENIAC, “occupied 167 square meters and consumed 150kW, the equivalent of 50 air conditioning units,” and that today it would take “400,000 ENIAC to have the same computing power as a mid-range smartphone, whose consumption is practically negligible.”
AI is also decisively contributing to environmental preservation. In fact, it is seen as a potentially key tool for the progress of the Sustainable Development Goals, as it achieves more results with fewer resources. López de Fez confirms that it is already being used to conserve biodiversity or manage waste. In terms of optimizing water use, for example, “it helps predict demand, identify leaks, and optimize distribution, contributing to more efficient management,” the engineer reviews.
This technology also allows for real-time data analysis and cross-referencing, making it an optimal solution for predicting natural disasters and creating alert and management solutions. It can also be a pillar for the sustainability of the primary sector. In fact, “it is being used to increase crop yields and minimize environmental impact,” explains López de Fez.
Thus, as has always happened when humans have been able to acquire a new tool, it remains to be seen how Artificial Intelligence will ultimately benefit those who have created it.
