From Tech to Justice: A Call for Environmental Justice in AI

Illustration by Somnath Bhatt

A guest post by Rachel Bergmann & Sonja Solomun. Rachel is a writer, researcher, and former research assistant with the Social Media Collective at Microsoft Research New England. Sonja is the Research Director of the Centre for Media, Technology and Democracy at McGill’s Max Bell School of Public Policy, a PhD Candidate at McGill University, and a Co-Founder of the Coalition for Critical Technology. Twitter: @rachbgm @SonjaSolomun.

This essay is part of our “AI Lexicon” project, a call for contributions to generate alternate narratives, positionalities, and understandings to the better known and widely circulated ways of talking about AI.


Tech companies and critical AI scholars have increasingly touted ‘AI sustainability’ and ‘Green AI’ as ways to address the energy demands from AI models and data-intensive systems, and to use these same systems to mitigate the climate crisis. These initiatives take many forms — from improving HVAC systems in the buildings and data centers of tech companies to prediction and risk management for agriculture, weather events, and climate disasters. Increasingly, many are turning to efforts to quantify, reduce, and recapture carbon. While these initiatives may be helpful in narrowly addressing specific and often exclusively technological impacts to the climate, they do little to address environmental justice — the indivisible integration of the environment with our broader social, political, and economic conditions. Despite clear overlap with broader concerns around power, bias, and discrimination, and with attending calls for fairness and accountability in relation to public and local contexts, sustainability remains largely siloed from emerging fields of research on AI ethics.

In this essay, we argue that our current methods of interrogating and understanding sustainability in AI are extremely limited. We aim to push beyond the current bounds of AI sustainability and “Green AI” by proposing environmental justice in critical AI discourse. Summarized by Schlosberg, Rickards, & Byrne as the “(in)equity in the distribution of environmental bads (and goods),” environmental justice opens up possibilities beyond flattening AI sustainability to “tech for climate” or abstracting it to global problems and universal solutions. While we will undoubtedly require global coordination to address the climate crisis, we need greater attention to how local AI infrastructures, including the vast networks of non-commercial facilities, global supply chains, and local workers who power them, continue to disproportionately impact communities of colour, extending long legacies of environmental racism today.

To this end, we offer environmental justice as a framework for rooting terms like “sustainable AI” and “Green AI” within their deep relationality and attachment to place. Anchoring environmental justice in AI within specific community and institutional practices situates it within, and helps reveal, intertwined lineages of oppressions. By calling for environmental justice in critical AI discourse, this essay aims to amplify scholarship which acknowledges and moves beyond differential impact and incorporates environmental justice principles, histories, epistemologies, and vocabularies into current critical AI discourse. If AI is to bring about the kinds of sustainability touted by tech companies’ public commitments, it must include humanistic, interpretive, race-conscious, and pluralistic representations of the Earth’s future in making climate change plans.¹

Sustainability and Green AI

The terms nature, environmentalism, and sustainability are notoriously wide-reaching and hard to define. Sustainability has been described as a “container term”² that includes things like development, environmental protection, finance, tourism, politics and a range of other goals. In policy, corporate, and development contexts, common definitions of sustainability and sustainable development include three pillars: economic, environmental, and social (also called the “triple bottom line”).³

In the context of AI research and practice, however, sustainability almost always implies some sort of environmental initiative. From “smart fisheries” to predictive modeling of climate change, ‘sustainable AI’ projects use computational methods to monitor and conserve the Earth’s ecosystems, as well as improve environmental resource management for cities, factories, supply chains, and buildings.⁴ Other projects include “smart cities” and sustainable urban development⁵; using sensors, drones, and computer vision to monitor wildlife and pollution⁶; and optimizing and “greening” global supply chains.⁷ By narrowly focusing on the intersection between technology and “the” environment, AI practitioners and scholars miss the opportunity to meaningfully engage across issues of economic and environmental justice, such as how labor rights, public health, and community agreements are constitutive of algorithmically mediated work for instance.

Current AI research engages very little with the social and economic aspects of sustainability. Even in research that does acknowledge sustainability’s social and economic pillars — for example, research on how AI could help implement the United Nations Sustainable Development Goals (SDGs)⁸ — we found little nuance or critical engagement with socially or economically just goals. Instead, the dominant literature tends to frame AI as a neutral and apolitical tool which will or will not universally solve a sustainable development goal like gender equality or food security. We found little engagement with the ways AI systems are used by people and institutions in particular geographic, cultural, and material contexts and which may help or hinder particular goals in complex and uneven ways.

Instead, current AI research on sustainability tends to emphasize the quantifiable effects of environmental pollution and climate change, and focus on solutions of continued measurement, monitoring, and optimizing for efficiency. Although projects like reducing wasteful energy usage are an extremely important aspect of environmental action, such a high-level view is quite abstracted from specific places and people who disproportionately — and presently — experience the burdens of environmental degradation, including communities living in California’s “Diesel Death Zones” or farmers competing with data storage centres for their water supply.

Scoping a sustainability project only to include quantitative, abstracted, large-scale measurements of energy usage or pollution makes it difficult to bring local political, historical, and ecological contexts and practices into account when defining environmental harms and potential paths forward. Below we outline how we envision incorporating environmental justice values and practices into mainstream AI research, in order to move beyond distributive justice frameworks.

From Sustainability to Environmental Justice in AI

We propose environmental justice as a new term to be incorporated into critical AI discourse, so that we can begin to critically interrogate the uneven power conditions through which the environmental harms of digital technologies are experienced. An anticolonial movement led in the US by Indigenous people, Black people, and people of colour, environmental justice is the collective response to the continual and disproportionate impact of environmental hazards on people of color, known as environmental racism.⁹ For Robert Bullard, often described as a “father of environmental justice” based in the American South, environmental justice seeks to “address power imbalances, lack of political enfranchisement, and to redirect resources so that we can create some healthy, liveable and sustainable types of models.”¹⁰ Growing out of the civil rights and Red Power movements of the 60s and 70s, the environmental justice movement was a unification across issues as well as a reaction to the race-blind, class-blind environmentalism of the 1970s.¹¹ As contemporary scholars like Kathryn Yusoff, Nancy Tuana and others have echoed, many manifestations of environmentalism, sustainability, and even the concept of the Anthropocene have been racially unaware and at times politically toothless.¹² At its core, environmental justice recognizes that “racial capitalism is not simply incidental to… our broader sustainability challenges. Instead it must be central to how we’re thinking about these challenges.”¹³

Unlike sustainable AI, environmental justice research centres justice, both to people and to nature, in its analyses.¹⁴ Scholarship in critical environmental justice moves beyond Rawlesian notions of distributive justice (e.g. the redistribution of rights, opportunities and material resources to address inequality), and instead builds justice frameworks based on feminist theorists like Iris Marion Young and Nancy Fraser, who argue that class, gender, sexuality and other categories of difference are inseperable from environmental exploitation.¹⁵ For instance, ​​the Center for Community Action and Environmental Justice (CCAEJ) was founded by a group of women living near — and successfully shutting down — the Stringfellow Acid Pit toxic waste site in the 1970s. As Brinda Sarathy documented, CCAEJ consciously mobilized based on locality, not on land-ownership, which enabled a legacy of multi-issue collective action across race and class (2013: 258). Today, CCAEJ members are continuing to bridge the artificial separation between environmental and economic justice in their fight against Amazon’s airport hub expansion, demonstrating how place, environment, worker status, and race are mutually constitutive as we’ve argued elsewhere.

Yet, this kind of relationality is missing from sustainable AI research, which often abstracts “the” environment not only from place, but from broader social, political, and economic conditions. But as Robert Bullard underscores, “The environment is everything: where we live, work, play, go to school, as well as the physical and natural world. And so we can’t separate the physical environment from the cultural environment. We have to talk about making sure that justice is integrated throughout all of the stuff that we do.”¹⁶ Rather than treating the environment and society as separate realms in AI discourse, environmental justice frameworks can foreground the “intrinsically dynamic” relations between them as Manuel Arias-Maldonado suggests.¹⁷

These relations receive little attention from cognate interdisciplinary fields of AI ethics, such as the ACM conference on Fairness, Accountability, and Transparancy (FAccT), which have mobilized a breadth of rigorous work and scholarship to address global patterns of AI discrimination and power. For instance, demands for ‘fairness’ have called for further justifications for the public use of AI systems beyond their technological efficiency. Instead, researchers and practitioners argue, fairness requires social evaluations and analyses of how AI systems are used in context, especially in high-stakes areas. Calls for greater transparency have asked regulators to move beyond ‘opening the black box’ of proprietary systems to instead disclose and make publicly available where AI systems are used, and for what purpose.¹⁸ Most recently, global emphasis on algorithmic accountability have coalesced around regulation and policy for decisions made with, and by AI systems, especially in the public-sector.¹⁹

Despite clear overlap, and especially given recent concerns about the environmental impacts of AI, sustainable AI currently exists siloed from FAccT discourse. At the time of this writing, FAccT currently has one paper that mentions sustainability, green, climate, or environmental impact.²⁰ Questions of environmental harm and injustice are not only important facets of algorithmic fairness and accountability, but environmental justice specifically could foreground both the slow, globally distributed violence of AI’s environmental harms as well as other social and economic issues like worker power, community empowerment, and democratic decision-making structures as inseparable from these environmental concerns.

Alternative Relations: Incorporating Environmental Justice

Building in relationality within environmental justice in critical AI discourse can bring in direct engagements with the histories and legacies of colonialism, settler colonialism, environmental racism, and normative “claims over nature” inherent in AI discourse from its early appellations.²¹ This may involve, for instance, including those most marginalized by technological interventions into the design process, or rethinking how we relate to machines more broadly.

In practice, environmental justice in critical AI research might move beyond computational efficiency as the measure of technological sustainability by requiring environmental justice assessments developed by those directly impacted; incorporating research that addresses environmental racism in AI infrastructures; requiring companies to adopt and integrate toolkits such as Race Forward’s Racial Equity Impact Assessments (REIA); demanding community benefits agreements in areas surrounding AI developmentals; or filing employee resolutions for companies like Amazon to “publicly acknowledge and reduce disproportionate environmental harms to communities of colour.”

At their core, such interventions bring history and place to bear on how AI systems support and maintain infrastructural and invisible violence in the global context. Rob Nixon calls this kind of violence slow violence — a violence which “occurs gradually and out of sight, a violence of delayed destruction that is dispersed across time and space, an attritional violence that is typically not viewed as violence at all.”²² The “Climates of Inequality” project for instance, is an excellent public history documenting the slow violence of industrial supply chains. An environmental justice lens helps critical AI scholars consider AI’s role in global systems of environmental, social, and economic exploitation while also grounding that violence in specific places and contexts.


Feminist scientist Jill Schneiderman notes the way in which slow violence leads to the naturalization of environmental degradation, which comes to be seen as inevitable.²³ Through a parallel double-blind, the “inevitability” of AI, we are told, will lead us out of the impending climate crisis. Foregrounding environmental justice to imagine sustainable AI futures makes clear that the climate crisis and environmental degradation are inseparable from the global and hyper-local contexts of historical, political, social, and economic injustices. Reframing sustainability and AI in terms of environmental justice offers a way to center the material contexts and implications of AI technologies and provides a framework for imagining community-led, socially just futures.

Many thanks to Luke Strathmann and Amba Kak for their thoughtful comments and advice on this piece.


[1] Castree et al (2014); “Climate Justice Alliance” (n.d.); “Grassroots Global Justice Alliance” (n.d.); “Indigenous Environmental Network” (n.d.); “The Red Black and Green New Deal: A National Black Climate Agenda” (n.d.).

[2] Vogt (2009), Reider (2010), quoted in Spindler (2013).

[3] Hacking & Guthrie (2007); Purvis et al (2019).

[4] Dauvergne (2020).

[5] Chui et al (2018); Quan et al (2019); Yigitcanlar & Cugurullo (2020).

[6] Honarmand Ebrahimi et al (2021); Gonzalez et al (2016); Ly et al (2019); Mrówczyńska et al (2019).

[7] Chin et al (2015); Sanders et al (2019).

[8] E.g., Vinuesa et al (2020).

[9] (1999).

[10] Bullard (1999) quoted in (1999).

[11] Cole & Forster (2001).

[12] MacGregor (2005, 2009); Tuana (2019); Yusoff (2019).

[13] Patterson et al (2019).

[14] Schlosberg (2009) distinguishes these two kinds of justice as environmental justice, concerned with environmental risks to human populations, and ecological justice, more concerned with justice to nature. Although they are sometimes separate, environmental justice projects like food security, climate justice, and indigenous environmental action bridge the gaps between environmental and ecological justice.

[15] Young (1990); Fraser (1995); Schlosberg (2009).

[16] Buller (1999) quoted in (1999).

[17] Arias-Maldonado (2013).

[18] Richardson et al. (2019).

[19] AI Now Institute (2021).

[20] Benami et al. (2021).

[21] Lewis et al. (2018); Amrute (2019).

[22] Nixon (2011:2).

[23] Schneiderman (2012)

Works Cited

AI Now Institute. “Algorithmic Accountability for the Public Sector: Learning from the First Wave of Policy Implementation.” Accessed June 8, 2021.

Amrute, Sareeta. “Of Techno-Ethics and Techno-Affects.” Feminist Review 123, no. 1 (November 1, 2019): 56–73.

Arias-Maldonado, Manuel. “Rethinking Sustainability in the Anthropocene.” Environmental Politics 22, no. 3 (May 2013): 428–46.

Asaka, Megan, Cathy Gudis, Rudolph Bielitz, Katrin Boniface, Yvonne Chamberlain, Emily Grucza, Emily Guerrero, et al. “Witnessing the Slow Violence of the Supply Chain.” Climates of Inequality (blog). Accessed March 26, 2021.

Benami, Elinor, Reid Whitaker, Vincent La, Hongjin Lin, Brandon R. Anderson, and Daniel E. Ho. “The Distributive Effects of Risk Prediction in Environmental Compliance: Algorithmic Design, Environmental Justice, and Public Policy.” In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 90–105. FAccT ’21. New York, NY, USA: Association for Computing Machinery, 2021.

Bergmann, Rachel, and Sonja Solomun. “Diesel Death Zones in the Amazon Empire: Environmental Justice in Algorithmically Mediated Work.” In This Seems to Work: Designing Technological Systems with the Algorithmic Imaginations of Those Who Labor. Yokohama, Japan: ACM, 2021.

Cabrera, Yvette. “Dr. Robert Bullard, the ‘father of Environmental Justice,’ Is Watching the Movement Grow Up.” Grist, March 23, 2021.

Castree, Noel, William M. Adams, John Barry, Daniel Brockington, Bram Büscher, Esteve Corbera, David Demeritt, et al. “Changing the Intellectual Climate.” Nature Climate Change 4, no. 9 (September 2014): 763–68.

Chin, Thoo Ai, Huam Hon Tat, and Zuraidah Sulaiman. “Green Supply Chain Management, Environmental Collaboration and Sustainability Performance.” Procedia CIRP 26 (2015): 695–99.

Chui, Kwok Tai, Miltiadis D. Lytras, and Anna Visvizi. “Energy Sustainability in Smart Cities: Artificial Intelligence, Smart Monitoring, and Optimization of Energy Consumption.” Energies 11, no. 11 (November 2018): 2869.

“Climate Justice Alliance.” Accessed June 8, 2021.

Cole, Luke W., and Sheila R. Foster. From the Ground Up: Environmental Racism and the Rise of the Environmental Justice Movement. 1st Edition. New York: NYU Press, 2001.

Dauvergne, Peter. AI in the Wild: Sustainability in the Age of Artificial Intelligence. The MIT Press, 2020. “Environmental Justice: An Interview with Robert Bullard.” Accessed June 8, 2021.

Fisher, Douglas H. “A Selected Summary of AI for Computational Sustainability.” In Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence, 4852–57, 2017.

Fraser, Nancy. “Recognition or Redistribution? A Critical Reading of Iris Young’s Justice and the Politics of Difference*.” Journal of Political Philosophy 3, no. 2 (1995): 166–80.

Gomes, Carla, Thomas Dietterich, Christopher Barrett, Jon Conrad, Bistra Dilkina, Stefano Ermon, Fei Fang, et al. “Computational Sustainability: Computing for a Better World and a Sustainable Future.” Communications of the ACM 62, no. 9 (August 21, 2019): 56–65.

Gonzalez, Luis, Glen Montes, Eduard Puig, Sandra Johnson, Kerrie Mengersen, and Kevin Gaston. “Unmanned Aerial Vehicles (UAVs) and Artificial Intelligence Revolutionizing Wildlife Monitoring and Conservation.” Sensors 16, no. 1 (January 14, 2016): 97.

Grassroots Global Justice Alliance. “Grassroots Global Justice Alliance.” Accessed June 8, 2021.

Hacking, Theo, and Peter Guthrie. “A Framework for Clarifying the Meaning of Triple Bottom-Line, Integrated, and Sustainability Assessment.” Environmental Impact Assessment Review 28, no. 2–3 (February 2008): 73–89.

Honarmand Ebrahimi, Sanaz, Marinus Ossewaarde, and Ariana Need. “Smart Fishery: A Systematic Review and Research Agenda for Sustainable Fisheries in the Age of AI.” Sustainability 13, no. 11 (May 27, 2021): 6037.

“Indigenous Environmental Network,” November 26, 2020.

Lewis, Jason Edward, Noelani Arista, Archer Pechawis, and Suzanne Kite. “Making Kin with the Machines.” Journal of Design and Science, July 16, 2018.

Ly, Hai-Bang, Lu Minh Le, Luong Van Phi, Viet-Hung Phan, Van Quan Tran, Binh Thai Pham, Tien-Thinh Le, and Sybil Derrible. “Development of an AI Model to Measure Traffic Air Pollution from Multisensor and Weather Data.” Sensors 19, no. 22 (November 13, 2019): 4941.

MacGregor, Sherilyn. “A Stranger Silence Still: The Need for Feminist Social Research on Climate Change.” The Sociological Review 57, no. 2_suppl (October 1, 2009): 124–40.

— — — . “No Sustainability without Justice: A Feminist Critique of Environmental Citizenship.” In Environmental Citizenship, edited by A Dobson and D Bell, 101–26. MIT Press, 2005.

Mrówczyńska, Maria, Małgorzata Sztubecka, Marta Skiba, Anna Bazan-Krzywoszańska, and Przemysław Bejga. “The Use of Artificial Intelligence as a Tool Supporting Sustainable Development Local Policy.” Sustainability 11, no. 15 (January 2019): 4199.

Nixon, Rob. Slow Violence and the Environmentalism of the Poor. Harvard University Press, 2010.

Purvis, Ben, Yong Mao, and Darren Robinson. “Three Pillars of Sustainability: In Search of Conceptual Origins.” Sustainability Science 14, no. 3 (May 1, 2019): 681–95.

Quan, Steven Jige, James Park, Athanassios Economou, and Sugie Lee. “Artificial Intelligence-Aided Design: Smart Design for Sustainable City Development.” Environment and Planning B Planning and Design 46 (October 1, 2019): 1581–99.

Racial Capitalism, Designs for Energy Transition and the Green New Deal. Transcription, 2019.

Richardson, Rashida, Jason M Schultz, and Vincent M Southerland. “LITIGATING ALGORITHMS 2019 US REPORT:,” n.d., 32.

Schlosberg, David. Defining Environmental Justice: Theories, Movements, and Nature. Oxford University Press, 2009.

Schneiderman, Jill. “Awake in the Anthropocene.” Contemporary Buddhism 13 (May 1, 2012).

Schwartz, Roy, Jesse Dodge, Noah A Smith, and Oren Etzioni. “Green AI.” Communications of the ACM63, no. 12 (December 2020): 54–63.

Spindler, Edmund A. “The History of Sustainability The Origins and Effects of a Popular Concept.” In Sustainability in Tourism: A Multidisciplinary Approach, edited by Ian Jenkins and Roland Schröder, 9–31. Wiesbaden: Springer Fachmedien, 2013.

“The Red Black and Green New Deal: A National Black Climate Agenda.” Accessed June 8, 2021.

Tuana, Nancy. “Climate Apartheid: The Forgetting of Race in the Anthropocene.” Critical Philosophy of Race 7, no. 1 (2019): 1–31.

Vinuesa, Ricardo, Hossein Azizpour, Iolanda Leite, Madeline Balaam, Virginia Dignum, Sami Domisch, Anna Felländer, Simone Daniela Langhans, Max Tegmark, and Francesco Fuso Nerini. “The Role of Artificial Intelligence in Achieving the Sustainable Development Goals.” Nature Communications 11, no. 1 (January 13, 2020): 233.

Wynsberghe, Aimee van. “Sustainable AI: AI for Sustainability and the Sustainability of AI.” AI and Ethics, February 26, 2021.

Yigitcanlar, Tan, and Federico Cugurullo. “The Sustainability of Artificial Intelligence: An Urbanistic Viewpoint from the Lens of Smart and Sustainable Cities.” Sustainability 12, no. 20 (2020): 1–24.

Young, Iris Marion. Justice and the Politics of Difference. Revised edition. Princeton, N.J: Princeton University Press, 1990.

Yusoff, Kathryn. A Billion Black Anthropocenes or None. Univ Of Minnesota Press, 2019.