🔬 Research summary by Sarah P. Grant, a freelance writer dedicated to covering the implications of AI and big data analytics.
[Original paper by Peter Dauvergne]
Overview: Industry leaders are quick to champion AI as a transformative force for environmental sustainability. In this research paper, Peter Dauvergne examines their claims through a critical international political economy (IPE) lens, and finds that AI’s environmental sustainability benefits for the global supply chain are overstated.
“We can’t save the world by playing by the rules, because the rules have to be changed.”
Delivering these words during her TEDxStockholm talk, Greta Thunberg made her position clear: that without tougher laws, the climate crisis can’t be tackled effectively.
But in an attempt to ward off tougher regulations, many transnational corporations (TNCs) are trying to promote self-governance, argues Peter Dauvergne in a paper on AI, global supply chains and environmental sustainability.
Here, Dauvergne draws on literature from international agencies, nonprofit watchdogs, journalists, business management consultants, and scholars to examine the issue from an IPE perspective. One of his major questions is whether AI’s eco-efficiency gains as promoted by TNCs will contribute to significant sustainability progress–and his answer is a definitive “no.”
He finds that AI is often used as a way to project an image of corporate social responsibility (CSR), but that this detracts from the issue of environmental harms posed by AI through impacts such as over-consumption, intensified resource extraction and e-Waste. He concludes that AI is not transforming TNCs into agents of environmental sustainability, and that CSR messaging can create a false sense of security, making it harder to govern industry players.
AI’s core business benefits
Dauvergne lays the foundation for his analysis by addressing the core business benefits of AI. He references survey data from consulting firm McKinsey to show how supply chain management is one of the major business activities where AI is yielding the most benefits. Manufacturers and retailers in particular stand to profit the most from the optimization of supply chains through machine learning and intelligent automation.
After delving into the business impacts of AI, he then examines how corporations have positioned their cost-cutting and efficiency activities as sustainable practices over the past 20 years. He observes that the value of new technology for solving global problems “has long been a key part of this CSR messaging. This is now the case with the framing of artificial intelligence as a key solution for achieving corporate sustainability.”
While Dauvergne’s main argument is that corporate metrics and CSR messaging are exaggerating the impacts of AI for environmental sustainability, he does acknowledge that many of the specific claims are true.
He states, for example, that the algorithms can enhance shipping by rerouting in case of bad weather and data-powered apps can cut emissions by reducing idling time for trucks. However, he asserts, these are really just micro-benefits that will only go so far when the main corporate purpose is sustainability in profit-making.
AI and its environmental costs
After focusing on how TNCs are highlighting the benefits of AI for environmental sustainability as part of branding, marketing and CSR messaging, Dauvergne then examines four major environmental costs of AI:
- Increased Energy and Mining Demands
Dauvergne argues that AI will be used to increase efficiency, and history shows that efficiency gains almost always lead to higher resource extraction, production and consumption. Firms from multiple industries promote AI as a greening technology that can be used to protect biodiversity, prevent tropical deforestation, or prevent carbon pollution. In reality, computers, smartphones, robots, data centers, are driving up demand for energy and mining.
- Over-Consumptionist Cultures
Dauvergne also emphasizes how advertising–critical to the business models of Google and Facebook–is leveraging AI to drive up consumption levels. He illustrates how deep learning techniques could increase the value of the packaged consumer and the global retail sector by turbocharging consumer demand and consumption.
- Impacts on Vulnerable Populations
Those who benefit from manufacturing outputs and those who are harmed by it are separated by geography, observes Dauvergne. AI is increasing demand for hardware that uses metal tantalum that is extracted from areas where human rights protections are minimal. The hardware also contains rare earth elements that create toxic waste, dumped in areas that are out of sight of populations living in wealthier states.
- Tsunamis of E-waste
The amount of e-waste is increasing substantially each year–and rose to “a weight equal to stacking up every commercial airliner ever built,” writes Dauvergne, referring to previous studies. Smart products are not designed to last, which will only contribute to the “tsunami of e-waste” that is “flooding the developing world with toxic chemicals and hazardous metals.”
AI sustainability: The political economy perspective
This research is an initial attempt to probe the environmental consequences of supercharging supply chains with AI. As Dauvergne notes in this paper, at the time that he conducted this research, almost no IPE scholarship had focused on this topic.
He provides several insights for future research and states that a more in-depth consideration is needed to examine how efficiency and productivity benefits can produce environmental harms. Dauvergne makes the case for research that does not take CSR messaging at face value.
Between the lines
This is an important article because it demonstrates that AI ethics is about more than simply executing technical fixes to make systems align with human goals. It’s also a matter of exposing what different groups in positions of power value, and how by helping them accomplish their goals, AI can in turn produce potentially harmful outcomes.
There is, however, another compelling area that future research could explore that is not covered in this paper. Here, Dauvergne focuses on the role of government as a regulator. In contrast, the economist Mariana Mazzucato has asserted in her book Mission Economy: A Moonshot Guide to Changing Capitalism that wicked problems like the climate crisis require a more flexible role for government, one in which it acts as a co-creator with the private sector–and not just as an institution that “fixes markets.”
Further research, then, could help address the question of whether AI’s potential environmental impacts would be mitigated by a more “mission-oriented” economic approach–one where the private sector’s goals are brought into alignment with a broader societal purpose.