From AI beer brewers to AI nurses for the elderly, there seems to be no limit to the potential of the relatively new technology of artificial intelligence. Although using AI may seem quite helpful to type up a quick email or brainstorm ideas for a presentation, scientists fear that AI is also further exacerbating the climate crisis through its energy consumption. For the past few decades, the world has strived to develop new strategies to mitigate climate change and its drastic effects; among these, renewable energy seems to be the most prominent. Despite the sustainable advances we have achieved through international agreements and legislation, AI’s exhaustion of energy makes it feel as though every step forward is met with three steps back.
Artificial intelligence is predicted to become one of the top sources of energy consumption worldwide; in 2022, it made up 2% of the globe’s total demand, however, this is expected to double by 2026—a development that could increase energy consumption by an amount comparable to the entire country of Sweden. Moreover, data centers, which host the technology for AI, are producing large amounts of electronic waste. Notably, they rely on scarce minerals, often mined in environmentally destructive manners, and consume copious amounts of water and energy. Research even suggests AI data centers consume six times more water annually than the entirety of Denmark, highlighting concerns over water accessibility long-term. This therefore raises the important question: can sustainability and artificial intelligence coexist?
Due to the vast environmental impacts of AI, it is important to acknowledge how it is sourced: thousands of data centers around the world are owned by large corporations like Google, Microsoft, Apple, and so forth. And, in terms of their accountability, the lines are blurry. These companies often fail to openly disclose how much of their energy consumption stems from AI operations compared to their regular energy consumption. The mere calculation of their carbon footprint and an extensive understanding of their environmental impact is often impossible to discern because of the inaccessibility of their energy consumption data. When their greenhouse gas (GHG) emissions are reported, they tend to be the direct ones, rather than the indirect ones—for example, all the energy used to cool data centers. This raises concerns as these large corporations often bypass governmental rule because of their international nature and their essential roles in countries’ economies. Are our leaders adapting fast enough to protect the environment in the face of the rise of artificial intelligence? Moreover, do they want to?
Though large intergovernmental organisations such as the European Union encourage the ethical use of AI, these recommendations are non-binding and therefore generally ineffective. Even the legislation introduced by some countries like the United States seems to be scarce, especially when it comes to the environment. These companies’ actions are becoming more obscure due to their untracked investments in different AI firms. For instance, Microsoft does not account for Open AI’s emissions despite their investments in these operations. The lack of standardized reporting of GHG emissions during AI operations and the financial prioritization of these large companies often results in the exploitation of loopholes to underreport the true extent to which natural resources are unsustainably consumed. This lack of accountability challenges AI regulation, putting the environment at stake and resulting in an indirect increase in these multinational corporations’ emissions with no resulting consequences.
Research has strongly emphasized the implementation of renewable energy to power up centers that utilize AI models and training, with the aim of becoming more sustainable. Yet, pragmatically, it seems impossible for green energy to catch up to AI’s developments. Data centers grew exponentially from 500,000 in 2012 to now around 8 million. And, despite these tech companies committing to green investments, their actions say otherwise.
Data centers take around one year to be constructed; meanwhile, several years are needed to construct the renewable energy infrastructure needed to power them. Implementing renewable energy inherently requires an eternal game of catch-up. Not to mention, these data centers are designed for traditional power grids; adapting this infrastructure to renewable energy would be an extra investment of time and money—a sacrifice these firms are not keen on making. The reliability of green energy is not guaranteed either, due to its dependence on climatic conditions and investors. As AI continues to develop and require more energy, green energy may be incapable of relaying the sufficient amount to meet the growing needs of data centers. This further portrays the complex relationship between AI and renewable energy–while it is necessary to invest in both, their relationship is subject to technological developments and governmental investment.
So, what’s next? It is evident AI consumption will continue to increase, arguably to the benefit of sustainable development. The United Nations Environmental Programme (UNEP) reports AI can play a critical role in managing the globe’s biggest environmental emergencies, as well as charting greenhouse gas emissions and optimizing logistics. However, in terms of corporate accountability to ensure sustainable development and environmental protection, there is a lot of work to be done.
Experts suggest ensuring the public disclosure of corporations’ investments’ GHG emissions to encourage accountability and awareness for stakeholders and clients alike. Furthermore, investing in the study of the creation of energy-efficient hardware and AI algorithms—through this optimization, one could guarantee increased efficiency and a simultaneous decrease in energy use. Governments are also encouraged to increase their agency by implementing strict standards and regulations to ensure the ethical usage of artificial intelligence on a national and international level. Research must be further prioritized to encourage collaboration between policy-makers, business owners, governments, and academics, duly resulting in comprehensive developments that incorporate a multi-disciplinary approach. Finally, fostering a positive relationship between AI and sustainability can encourage its use as a tool for environmental protection and the achievement of the UN’s Sustainable Development Goals. Though AI is not perfect, it can be molded into a valuable tool, rather than a burden on humanity’s development.
Other posts that may interest you:
Discover more from The Sundial Press
Subscribe to get the latest posts sent to your email.
