Executive summary

With AI and technology being more accessible, more carbon footprints are produced. Though considering the benefits of technology in an increasingly interconnected world and greater scientific power, the production of this technology is in direct relation to poor environmental consequences. Considering the usage of technology to reverse poor climate impacts, this brief will deconstruct the impacts of technology in pursuit of greater social and scientific advancements. 

Overview

Though the rapid development of AI has brought about increased change and advancements, its power can have detrimental environmental effects. AI has been known, in the past decade, for its increased support in combating climate detriments. From predicting natural disasters, which have the potential to rescue millions of citizens, and leveraging greater existing data, it has been proven to be a powerful tool in reversing the damages of climate change.  However, it appears the downsides outweigh any inherent benefits. With its massive carbon footprint usage and environmental impact, increased digital usage poses a new question if its work outweighs its carbon damage.

History

Since the 1970s, the growth of electronic devices has only grown exponentially. As they became increasingly popular, concerns about its sustainability and overall environmental impact became heavily controversial and led to the rise of climate change activists. Though there have been ironically more technology usage to combat its environmental damage, its impending long term effects are still examined. 

Electronic waste, or e-waste, first appeared in landfills and runoff streams as a result of irresponsible disposition of digital technologies. The lack of recycling led to an excess buildup of hazardous materials, including mercury, lead, and cadmium in natural land and water bodies, impacting the environmental health of areas surrounding cities. These metals, which were incredibly laborious and intensive to extract and refine, made the production of digital technologies demanding, leading to increased greenhouse gas emissions, resource depletion, and land degradation, ultimately leading to rapid climate change. 

Energy consumption was also an extreme concern of digital technology critics. According to the Geneva Environment Network, energy consumption increased by almost 70% between 2013 and 2020. As the world becomes more dependent on digital technologies, the issue will only get more concerning. As of now, digital technology makes up the majority of our day-to-day tasks, communication, entertainment – ranging from GPS systems to e-commerce. As the digital world gets more complex and the world becomes more dependent on it, its impact on the environment will compound exponentially, and ultimately lead to even more destructive climate change and global warming. Greenhouse gases, one of the largest contributors to global climate change, is a topic of conversation within the digital technology sector, and steps are being made to mitigate the issue. On a worldwide scale, over 1.2 trillion grams of CO2 are produced every day just from email usage. Compounded with all the other available platforms available on the digital stage today, CO2 production has become a grave point of distress to climate change activists. 

Some prior actions have been taken to address this issue of energy consumption and e-waste. One such example is the Circular Economy Action Plan in the European Union, which was created to increase the sustainability of digital products through processes such as reusing, recycling, and refurbishing. By doing this, consumption became more sustainable, and e-waste and pressure for natural resources were significantly reduced. Another example of past actions include the Green IT policies, which had the goal of reducing emissions, waste, cost, and consumption of digital technology. To do this, these policies utilized four techniques that would ultimately reduce e-waste: disposition of technology responsibly, recycling unrepairable electronics, reselling individual components, and refurbishing and reusing the working devices. 

While digital technology has contributed to the increased worldwide communication and entertainment that we have today, its environmental burden is immense. Continued efforts to mitigate the issue through sustainable practices is necessary to the growth of technology in the digital age.

Background

In only two decades, digital technology has reached around 50 percent of the developing world's population, and has played an integral role in enhancing connectivity, financial inclusion, and access to trade and public services. These technologies and their equipment, however, directly contribute to a major ecological problem – digital pollution.

‍Digital pollution is the carbon emissions behind our digital activity and, as of 2019, is responsible for 4% of global greenhouse gas (GHG) emissions which is 50% more than the GHG emissions of air transport (2.4%). The use of digital technologies also accounts for 4.2% of primary energy consumption and 0.2% of water consumption worldwide. Digital technology’s share of global GHG emissions is rising sharply and could double by 2025 to reach 8%, according to the Shift Project’s report on the environmental impact of digital technology and 5G rollout. The environmental impact of digital technologies, however, is not limited to greenhouse gas emissions. The digital sector is also dependent on a growing number of critical resources. In the study ‘Towards Digital Sobriety’, written by Frédéric Bordage, a European-based digital expert in sustainable technology, the major components needed to power digital infrastructure are around 75 million servers, 34 billion User terminals, and 1.5 billion communication networks. With these numbers predicted to grow 5 times between 2010 and 2025, this colossal amount of equipment will require gradually more raw materials, resources and energy to continue expanding. 

In addition to concerns about unsustainable practices, there is also the social issue of how these raw materials are sourced and its environmental impact. A non-negligible proportion of the supply comes from illegal child labour operations. In the Democratic Republic of Congo, one of the main producers of Tantalum, armed conflicts over the exploitation of natural resources have been raging for 20 years. The sourcing of these raw materials, through ore extraction and element separation, have also led to the generation of chemical effluents (cyanide, arsenic, lead, sulphates, mercury, etc.) which are harmful to biodiversity and human health. In addition, open-pit mining alters the landscape, the soil, and the local hydrological regime.

The rapid advancement of AI, and its carbon footprint has also contributed to the environmental impact of technology. Carbon footprints assess the amount of carbon-producing energy used to do something. The carbon problem associated with the AI industry primarily revolves around two crucial stages: training and deployment. The training of AI models requires vast amounts of energy and emits substantial carbon emissions. For example, the training of GPT-3, one of the models, resulted in 552 metric tons of carbon emissions, equivalent to driving a passenger vehicle for over 2 million kilometres. The deployment of AI also raises concerns about collective energy consumption resulting from AI’s widespread use. The energy consumption of AI models taking on real-world tasks, such as ChatGPT, has sparked justified concerns about its environmental impact.

Policy Problem

As innovative technologies such as artificial intelligence and quantum computing continue to rise in their popularity and usage worldwide, many environmental critics and scientific experts have started to raise critical concerns surrounding the negative transformative impact that the development and deployment of such technologies could cause on ecosystems around the globe. The global market for AI technologies has tremendously grown in recent years, with reports showing that these services generated a revenue of about $62.5 billion dollars in 2022, an immense increase from the $19.4 billion in 2019. As a result, computing needs have been ever-rising in order to fuel various processes behind the implementation of such technology, including data preprocessing, training, inference, testing, and more. 

A critical example of this is software company NVIDIA which specializes in AI computing where they are estimated to manufacture and distribute 1.5 million AI server units annually by the year of 2027, which experts say can consume at least “85.4 terawatt-hours of electricity annually—more than what many small countries use in a year”. Such an immense amount of electricity has been claimed to be vital for fueling the intense computational power that are needed by Graphics Processing Units (GPUs) employed by data centers worldwide. This immense computing power has proved necessary due to the rapidly increasing complexity and size of huge datasets that are used to train large-scale AI models, along with the billions of parameters that are now being utilized within them to make the models more accurate and efficient. It has been estimated that around nine to 11 thousand cloud data centers have been constructed for these purposes.  Complemented by the rise of hyperscale data centers, which occupy around 1 to 2 million square feet of land compared to 100,000 square feet that is taken up by a traditional data center, it further intensifies their negative harms.

As a result of the energy-hungry GPUs that are used within these data centers, greenhouse gas emissions from them have rapidly increased as they continue to expand, leading to concerns surrounding air pollution, global warming, and climate change. For example, OpenAI was estimated to produce around 500 metric tonnes of carbon dioxide from its data centers when it was training ChatGPT, its state-of-the art Large Language Model (LLM) that has now made it world-renowned leader within the field of artificial intelligence and machine learning. These data centers also require an immense amount of electricity to fuel their operations through powering up their electronics, which has led them to consume up to 1 to 1.5% of the global electricity supply, according to a report commissioned by the International Energy Agency. Furthermore, the immense amount of electronics within these centers requires extensive amounts of water to cool them down and ensure that they are able to properly operate without excessive overheating. This has caused Google to increase its water usage by 20% within its data centers year-to-year from 2021 to 2022, while Microsoft’s increased by 34% during the same period of time. This can have especially negative effects on the geographical regions they are located in, with many of them being drought-prone, raising significant concerns for civilians residing in these areas about their water supply. For example, Google was shown to “consume more than a quarter of all the water used” in the city of Dallas, Oregon, exemplifying the need for more transparency from these corporations to ensure that essential water supply is not diverted from civilian use for such purposes.

Policy Options

The rise of artificial intelligence unequivocally has transformed numerous aspects of our society and will continue to have drastic impacts in the future. No one can doubt the implications for innovation, efficiency, and convenience across various mediums that AI offers. Unfortunately, even the best things have their downsides. The environmental impact of developing AI technologies cannot be overlooked. From the manufacturing processes of hardware components to the amount of rare earth minerals required to synthesize AI, to even the sheer energy consumption of data centers powering AI algorithms, the environmental footprint of AI development is substantial. Considering the significant nuance related to  AI and how indisputable it is that this tool is helping human life, shutting down AI development for environmental reasons should be out of the question. Should we place restrictions on AI development in response to its environmental impact, but at the cost of hindering the advancement of new technologies? Or, should we ignore its environmental impact and continue with the progression of AI at the current pace? Here are some policy options that look to achieve the best of both worlds:

A.  Federal Government mandate for AI companies to transition to renewable energy by 2030:

This is one of the more forward approaches when it comes to reducing the environmental impact of AI, but arguably the most effective. This mandate can be supported by some level of government subsidization, especially for smaller tech start-ups but should overall encourage companies to do so autonomously. Not only would this mandate provide greater innovation in the world of renewable energy, but it would effectively eliminate the use of fossil fuels within the AI development process. This option essentially holds the US market hostage against all massive and small tech companies. Knowing that the US is the most desirable market for AI consumption Tech companies will most likely follow the mandate leading to cleaner practices in AI manufacturing. This mandate could potentially have a snowball effect in other fields as well: a company being forced to ensure one sector of their energy consumption comes from a renewable source might be more inclined to make the change entirely. Overall, pressure from the Government might be exactly what we need to lower the environmental impact of such a growing industry like AI development. This policy also works to achieve the best of both worlds because although it is placing restrictions on AI development, the 6-year window allows AI to continue to develop while positive change is happening. 

B. Promotion of Energy-Efficient AI Technologies: 

This option takes the opposite approach of the mandate. Instead of attacking where AI gets its energy from, attack how much energy AI uses. This approach is arguably more sustainable and better in the long term than the one previously mentioned. Utilizing this approach will lead to AI now and in the future to require less energy, leading to a more environmentally sound practice. This could potentially turn the sources of AI’s energy obsolete due to the significant decrease of how much is required. Essentially, under the ideology of making AI more energy-efficient, a tech company can source their energy from wherever they want and the environmental impact will still be lowered due to the sheer fact the new AI requires less energy to operate/test. This can be done through implementing energy-efficient standards for AI hardware components, such as processors and GPUs, to encourage the development and adoption of energy-efficient technologies. The government could provide incentives, such as tax breaks or grants, for research and development focused on optimizing algorithms and testing procedures for reduced energy consumption. Overall, attacking the intersection between energy consumption and AI from either direction will hold significant benefits for humanity in the long run and allow AI production to continue in a more sustainable way.

By implementing either of these policy options, we can mitigate the environmental impact of AI development while fostering continued innovation and progress in the necessary field of AI. It is crucial to acknowledge the importance of addressing environmental concerns without hindering the advancement of AI technologies crucial for the advancement of the human race. 

Next Steps/Conclusion

As our planet faces climate change, there are a multitude of ways that technology has harmed and can potentially help with this issue. The topic of how energy consumption could be influencing the planet's climate inevitably arises throughout discussions of energy use. These advances are the result of innovation policies as well as economic and environmental concerns. However, implementing such technology is much easier proposed than done. 

Technological advancements have had a longstanding effect on the world’s environment. As technology becomes a stronger pillar of our modern-day lives, we must continue to find a technological and environmental policy as a means to mitigate pressing matters our environment is facing.

Acknowledgment

The Institute for Youth in Policy wishes to acknowledge Michelle Liou, Joy Park, Nolan Ezzet and other contributors for developing and maintaining the Policy Department within the Institute.

References

1. “Circular economy action plan - European Commission.” n.d. Environment.ec.europa.eu. 
2. https://environment.ec.europa.eu/strategy/circular-economy-action-plan_en.
3. “Data, Digital Technology, and the Environment – Geneva Environment Network.” 2024. Geneva Environment Network.
4. https://www.genevaenvironmentnetwork.org/resources/updates/data-digital-technology-and-the-environment/.
5. “The Environmental Impact of Digital Technologies and Data.” n.d. DataCamp. Accessed May 20, 2024. 
6. https://www.datacamp.com/blog/environmental-impact-data-digital-technology.
7. Lutkevich, Ben. n.d. “What is green IT (green information technology) and why is it important?” TechTarget. Accessed May 20, 2024.
8. https://www.techtarget.com/searchcio/definition/green-IT-green-information-technology.
9. Acaroglu, Leyla. 2024. “The Hidden Environmental Impacts of AI.” Wikipedia. 
10. https://www.linkedin.com/pulse/hidden-environmental-impacts-ai-leyla-acaroglu-xryyc/.
11. “AI's impact on the environment.” 2023. Digital Watch Observatory. https://dig.watch/updates/ais-impact-on-environment.
12. “Digital sustainability.” n.d. Iberdrola. Accessed May 20, 2024. https://www.iberdrola.com/sustainability/digital-pollution.
13. Engérant, Luce, Nathan Barrieu, Hélène Chauviré, and Mélodie Pitre. 2023. “Materials of the dematerialized: are there raw material supply
14. risks for companies in the digital sector ?” Carbone 4. https://www.carbone4.com/en/analysis-risk-raw-material-digital.
15. “The Impact of Digital Technologies | United Nations.” n.d. the United Nations. Accessed May 20, 2024.
16. https://www.un.org/en/un75/impact-digital-technologies
17. McNutt, Louise. 2023. “What is the environmental impact of digital technology?” Wedia.
18. https://www.wedia-group.com/blog/what-is-the-environmental-impact-of-digital-technology.
19. Violino, Bob. 2023. “Much of the tech world's climate impact is still being overlooked, and it's accelerating.” CNBC.
20. https://www.cnbc.com/2023/09/09/much-of-tech-world-climate-impact-is-overlooked-and-its-growing.html.

21. “What is Digital Pollution and easy actions to reduce our digital carbon footprint.” 2022. Business in the Community Ireland.
22. https://www.bitc.ie/newsroom/news/digital-pollution/.
23. Leffer, Lauren. “The AI Boom Could Use a Shocking Amount of Electricity.” Scientific American, February 20, 2024. https://www.scientificamerican.com/article/the-ai-boom-could-use-a-shocking-amount-of-electricity/.
24. Berreby, David. “The Growing Environmental Footprint of Generative AI.” Undark Magazine, February 21, 2024. https://undark.org/2024/02/20/ai-environmental-footprint/?gad_source=1&gclid=Cj0KCQjw6auyBhDzARIsALIo6v-9fDdYwYa1x9d45juRL5TPLA-DOIdqJxP6fRTHGuYa-ANwPhthhOAaAqcbEALw_wcB.
25. Coleman, Jude. “AI’s Climate Impact Goes Beyond Its Emissions.” Scientific American, February 20, 2024. https://www.scientificamerican.com/article/ais-climate-impact-goes-beyond-its-emissions/.
26. IEA. “Data Centres & Networks - IEA,” n.d. https://www.iea.org/energy-system/buildings/data-centres-and-data-transmission-networks#overview.
27. Fortune. “A.I. Tools Fueled A 34% Spike in Microsoft’s Water Consumption, And One City With Its Data Centers Is Concerned About The Effect on Residential Supply,” September 9, 2023. https://fortune.com/2023/09/09/ai-chatgpt-usage-fuels-spike-in-microsoft-water-consumption/.
28. Pascual, Manuel G., Manuel G. Pascual, and Manuel G. Pascual. “Artificial Intelligence Guzzles Billions of Liters of Water.” EL PAÍS English, November 15, 2023. https://english.elpais.com/technology/2023-11-15/artificial-intelligence-guzzles-billions-of-liters-of-water.html
29. Young, Chris. “Oregon'S City of the Dalles Agrees to Reveal Google’s Local Water Usage.” The Reporters Committee for Freedom of the Press, March 29, 2023. https://www.rcfp.org/dalles-google-oregonian-settlement/
30. Krasadakis, G. (2023, October 20). The Impact of Artificial Intelligence on the Society and Everyday Life | 60 Leaders. Medium. https://medium.com/60-leaders/the-impact-of-ai-on-society-and-everyday-life-711307e06b87#:~:text=AI%20is%20changing%20the%20way,help%20 humans%20make%20 better%20 decisions.
31. Perry, A. M., Keller, J. B., & Donoghoe, M. (2024, January 29). The US must balance climate justice challenges in the era of artificial intelligence. Brookings. https://www.brookings.edu/articles/the-us-must-balance-climate-justice-challenges-in-the-era-of-artificial-intelligence/#:~:text=Despite%20their%20 operation%20in%20a,form%2C%20through%20 rising%20energy%20 demands.
32. Coleman, J. (2024, February 20). AI’s Climate Impact Goes beyond Its Emissions. Scientific American. https://www.scientificamerican.com/article/ais-climate-impact-goes-beyond-its-emissions/#:~:text=But%20as%20 AI%27s%20 popularity%20 keeps,way%20AI%20 affects%20the%20 climate.
33. Williams, R. (2024, January 5). The Download: producing rare earth minerals, and future AI regulation. MIT Technology Review. https://www.technologyreview.com/2024/01/05/1086217/the-download-producing-rare-earth-minerals-and-future-ai-regulation/
34. Vincent, J. (2024, February 16). How much electricity do AI generators consume? The Verge. https://www.theverge.com/24066646/ai-electricity-energy-watts-generative-consumption
35. DeGeurin, M. (2024, March 25). AI companies eye fossil fuels to meet booming energy demand. Popular Science. https://www.popsci.com/technology/ai-power/
36. Artificial Intelligence - Global | Statista Market Forecast. (n.d.). Statista. https://www.statista.com/outlook/tmo/artificial-intelligence/worldwide#:~:text=Artificial%20Intelligence%20%2D%20Worldwide&text=In%20global%20comparison%2C%20the%20largest,US%2450.16bn%20in%202024).
37. “Evidence - NASA Science.” NASA. Accessed May 16, 2024. https://science.nasa.gov/climate-change/evidence/ 
38. Pipa, Anthony F., Tedros Adhanom-Ghebreyesus, and Adie Tomer Jenny Schuetz. “Cutting through Environmental Issues: Technology as a Double-Edged Sword.” Brookings, July 28, 2016. https://www.brookings.edu/articles/cutting-through-environmental-issues-technology-as-a-double-edged-sword/