The world is transitioning into a data-driven era, where algorithms can influence anything from our health-care services, consumer behaviour, and even national and international environmental decisions. Fears around the embedded bias and discrimination seen in algorithms has motivated an Environmental Data Justice movement of citizens, researchers, and technical practitioners who aim to make data more equitable.

Growth of data

As we navigate the climate crisis, we find that the scale of the challenges we face, like the ecosystems they threaten, are vast. One way researchers aim to deal with this scale and complexity is through the development and application of powerful computational techniques that can tap into underutilised and highly diverse environmental data sets. From finding better ways to manage resources, forecasting future crises, and tracking progress against the Sustainable Development Goals, a growing proportion of “the next generation of environmental scientists are data scientists”.

With big corporations advocating for the continued growth of such data, energised by the prospect of conquering a greater percentage of “untapped” sources such as imagery, social media feeds, emails, journals, and videos, the echoes of our Western history of “exploration”, “discovery”, and empire should not be ignored. With a growing abundance of data, the real task is often knowing how best to use it. But as many, emboldened by notions like innovation and breakthrough, focus on efficiency, is there enough thought spared for equality?

As data becomes ever more pervasive in our lives and instances of prejudice and discrimination as a result of technology become more frequent, there is growing pressure from a community of researchers, activists, and artificial intelligence (AI) practitioners to make Environmental Data Justice (EDJ) a top priority. Motivated by the capacity for data, much of which is generated by companies and governments, to perpetuate social norms and status quo that often share roots in oppression, researchers in the EDJ field aim to create future technologies that enable greater wellbeing, with the goal of beneficence and justice for all. One overarching and fundamental concern in the data justice field is the ability of data and AI practitioners to decide what and whose knowledge and data is counted as valid, and what goes ignored and unquestioned, identifying it as a form of power that should not continue to operate without critique.

Data's downfalls

Introduced by the Environmental Data and Governance Initiative (EDGI) in the aftermath of the 2016 US Election, the concept of EDJ has been implemented through a variety of archival and action based techniques. For Lourdes Vera, a PhD Candidate in Sociology at Northeastern University and EDGI member, the EDJ framework, “exists to challenge the ‘extractive logic’ of current federal environmental policy and wider data infrastructures”. On the topic of extractive logic, Lourdes turns to the words of Frantz Fanon's Wretched of the Earth. Fanon highlights how colonialism depersonalises the individual and this depersonalisation is felt in “the collective sphere”. It is in this space, for Lourdes, that extractive logic lives—it is just another part of the “legacy of capitalist, colonial logic”.

In one sadly typical research project, briefed by a large, multinational corporation “developing a sensor network for the elderly so that they could remain independent in their homes as they aged”, markets were targeted based on four main “cultural” groups, chosen for their similar economic status: Brazil, Russia, India, and China. It would be assumed that product research intended for deployment in such a range of communities and countries would be conducted in those countries, but with “little money available for travel”, the research was conducted in American suburbs. Perpetuating behaviours of coloniality by “rooting” cultural categories “in geographical separation” the research, as an example treated “Chinese-Americans as proxies for Chinese populations”, and used “diasporic communities [as] substitute for studies of cultures in a pure or home context”. “Culture” in these methods was taxonomised, and the identity of citizens compromised at the expense of a budget.

Power dynamics, in addition to coloniality, are also important in EDJ conversations. Environmental data injustices are, to Jennifer Gabrys, Chair in Media, Culture and Environment in the Department of Sociology at the University of Cambridge and Principal Investigator on the European Research Council (ERC) project Citizen Sense, “structural in the sense of the infrastructures in place for people to be able to set up data systems, harvest value and in who becomes an unwitting victim”. For Gabrys, it is necessary for the scientific community to look critically at the field; “we must ask ourselves who is actually able to generate different insights and profits from extracted data and understand that instances of data oppression neatly map on existing systems of privilege and inequality”.

Last autumn the US saw this exact description of data oppression play out in real life in a “widely used” US health-care algorithm where Black patients were systematically under-diagnosed and received lower health risk scores than their white counterparts. Why? The algorithm used health-care spending as a proxy for medical needs, but income is a good predicator for race in the US, with Black people being largely represented in lower socioeconomic groups. Additionally, Black patients receive a lower level of care in the health-care system, resulting in fewer visits to health-care services and lower usage of health insurance. Yet, applied science, especially in the environmental and health fields, is not only a product of epistemic values, but also of contextual values that reflect moral, societal or personal concerns in the application of scientific knowledge, and the absence of contextual considerations in this scenario perpetuated inequality and injustice.

As industry and governments increasingly look to AI practitioners and researchers for the solutions to important societal issues, understanding the systemic and permeative nature of racism, discrimination, and prejudice in our academic spheres is imperative to appreciating the real danger posed by unethical artificial intelligence and data practices. The recruitment, experiences, and opportunities of and for the BIPOC (Black, Indigenous, and People of Colour), LGBTQIA2S+ community, and other marginalised communities in academic and corporate spaces directly influences the diversity of the data field—any biases or discrepancies within academia will be propagated to other places that AI touches and feed back into the next generation of academics where it may further enhance systemic biases. These effects, it seems are compounded by the nature of algorithms which embody the politics of their design and development, replicating the tropes, biases (conscious or unconscious), and prejudices of the individuals creating and controlling them.

The future of data

The importance of lived experience in conversations around environmental data may seem frivolous, but for the BIPOC communities bearing the brunt of both environmental and data injustices, the intersections are undeniable. Through her research on how data visualisation can help communities mobilise to address environmental health and justice concerns, Lourdes Vera has shown, and continues to see, AI being used by state agencies like the US Environmental Protection Agency (EPA) to decide which toxic or polluting facilities will be dealt with in regards to non-compliance and violation of environmental regulations. The algorithms exist to support teams with limited resources, but the “blind trust” in algorithms to make models and decisions on polluting facilities without any community input or environmental justice consulting, alongside an “inbuilt colour blindness”, for Vera, poses a very dangerous reality. The questions that arise for her are: “what data is included, what data is ignored, and what knowledge systems and perspectives influence and lead these decisions?”

As Jennifer Gabrys puts it, “having a plurality of systematic approaches and ontologies is really important in capturing other experiences”, but one of the major issues in the field of data science, is there commonly being only “one accepted way of collecting data, monitoring problems, and designing studies”. The Citizen Sense project sets out to contextualize, question, and expand on the understandings and possibilities of democratised environmental action through citizen sensing practices.

In Deptford, southeast London for example, Citizen Sense found that pollution became part of wider housing injustice through dispossession due to gentrification, and inadequate transport infrastructure. Cognizant of the influences of systematic neglect that intersect with environmental struggles, Gabrys notes that it's not simply a matter of setting up a sensor and extracting data, but that “it is an ongoing process that requires attending to many different forms of justice, economic, environmental, social, and racial... it's also about the processes of governance, whose voice and evidence is counted and accepted”. Lourdes' work embodies these concerns, focusing on combining community engagement with empiric observations to give the communities she works with, “the ability to receive results, share them with neighbours and ignite community mobilisation”, which she observes is just as important as the raw data.

While achieving environmental data justice seems challenging in practice, particularly due to the extractive business models employed by existing tech giants, there is growing momentum behind more decentralised, community-led methods of gathering, and analysing data. The challenge will be developing these methods in ways which contend with existing infrastructures and shift the landscape towards greater justice.

About the author

Joycelyn Longdon is a Cambridge PhD Student on the Artificial Intelligence for Environmental Risk programme and Founder of ClimateInColour, a platform at the intersection of climate science and social justice. As a diasporic woman of color, Joycelyn says she cannot see climate justice without racial and social justice. Her work in the tech and science space focuses on centering indigenous knowledge systems and marginalized voices in algorithms.