October 3, 2024

Between September 2023 and September 2024, Data & Society researchers:

• Published 25 peer-reviewed journal articles and academic book chapters, 

• Held key leadership positions in ACM’s 2024 FAccT conference,

• Led three panels at the EASST-4S 2024 conference in Amsterdam,

• Edited a special issue of First Monday,

• Published 11 D&S reports, primers, and briefs.

As we mark Data & Society’s tenth anniversary this year, we’re reflecting on the organization’s earliest days and our particular style of social science research. It has always been important that our researchers produce — and that D&S publishes — original work focused on actionable empirical evidence. But we’ve also understood the importance of contributing to the larger field through traditional academic means: that is, peer-reviewed work in journals, books, and conferences. These are vital outlets for new knowledge and important venues for our researchers to build credibility and connections. 

Today, our researchers are just as academically active as ever. Over the last year they published at least 25 journal articles and academic book chapters. Former Director of Research Jenna Burrell and Program Director Jacob Metcalf worked together to co-edit a special volume of First Monday, collecting work on the “ideologies of AI and the consolidation of power.” And our researchers were instrumental in organizing this year’s conference on Fairness, Accountability, and Transparency (FAccT): Meg Young served as general co-chair and planning committee member, and Emnet Tafesse as DEI representative.

Across all of this work, familiar themes emerged: Too often, people want to treat sociotechnical problems as purely technical, and technologies are too often used for the consolidation of power above any other goal. In every case, taking a hard look at reality is critical to guarding our collective future. While artificial intelligence (AI) garnered much public attention, for example, our researchers were more focused on refining robust accountability methods than wrestling with the hype of generative systems. We also clearly saw the growing importance of climate-focused work, including research by AIMLab Project Director Tamara Kneese documenting how the professional practice around data-heavy work can fall short of real climate or ecological improvements. Workplace surveillance remained a perennial concern, with Program Director Aiha Nguyen and Researcher Alexandra Mateescu continuing to push questions of datafied labor into new professional sectors and new international contexts.

Algorithms, AI, and Accountability
D&S researchers have been studying so-called AI for years; their recent efforts have focused on creating meaningful accountability methods. AIMLab researchers have begun to take on concrete case studies, examining what goes right and wrong in practice. The team worked with partners at Cornell University and the Ada Lovelace Institute on a pair of FAccT papers — “Auditing Work” and “Null Compliance” — based on an examination of NYC’s ground-breaking AI-in-hiring accountability law, and both showed how the law ultimately falls short of its goals. (The former even won “Best Paper” in its category.) 

The question of what makes an effective algorithmic audit is far from settled. Researcher Briana Vecchione contributed to a pair of articles providing a taxonomy of the wide range of AI audit practices. In doing so, these articles discover the conditions necessary for an effective AI audit — conditions that are, unfortunately, rarely met. Meanwhile, Metcalf and Senior Researcher Ranjit Singh took up the question of whether “red-teaming” (the simulation of adversarial technology use) can adequately address the risks inherent in large language models by harnessing the collective “mischievousness” of a diverse group of users.

Running beneath these questions of assessment is a foundational dilemma of participation. As Young considers with her co-authors, what does effective participation actually look like in the context of new foundation models? Following this question to its logical end is one of the main directions of our ongoing research.

Members of our Trustworthy Infrastructures team also tackled the question of AI: researcher Tiara Roxanne co-authored an article with Mimi Ọnụọha on artistic and experimental approaches in AI anarchies, while former Director of Research Sareeta Amrute tackled feminist approaches to AI, grounded in affective experience, particularly those of people for whom AI systems do not work. 

Infrastructural Struggles
Physical and technical infrastructures can have huge impacts on human lives, even as they can seem invisible. This was an underlying theme of one of our 4S panels in 2024, where researchers Livia Garofalo and Joan Mukogosi presented on how new digital infrastructures upend traditional concepts of expertise — often in the settings of care and care work. In addition, a pair of D&S researchers, Ranjit Singh and Tamara Kneese, have each explored how the decisions surrounding infrastructures codify the distribution of power. Singh bases his work in large part on earlier dissertation research on India’s Aadhaar system, the national digital ID system meant to administer their modern welfare state. From there, his work has grown to examine ever more dimensions of infrastructure — their uneven distribution, the need to “slow” them down, and the ways they rely on intermediaries to work as designed. 

Kneese’s work comes at data infrastructures from another direction, often contextualizing them against the looming climate crisis. She asks, for instance, about how unexpected demands of infrastructure — like those required for archiving media from those who’ve died — present challenges for making “cloud” technologies sustainable. Kneese and fellow researcher Young also write on the difficulties of producing accuracy in assessments of data infrastructures for AI, accuracy that is required if we expect to actually have a positive effect on climate change.

The Surveillance of Labor
Our labor-focused researchers have written about workplace surveillance for years, often focusing on how new technologies enable and/or mask data collection. This year, Program Director Aiha Nguyen summarized how the new digital economy has led to pervasive surveillance and data collection, misclassification, a degradation of workplace standards, and increasing precarity for workers.

Both Nguyen and Labor Futures Researcher Alexandra Mateescu focus on how small acts of surveillance can add up to major imbalances in power. Nguyen and her co-author Wilneida Negrón explore how “little tech,” used to complete daily tasks, harms workers by increasing the risk for discrimination against employees. Similarly, Mateescu’s article on Electronic Visit Verification, which uses GPS tracking and biometric identity verification for homecare workers, argues that digital surveillance enables Medicaid bureaucracies to engage in continuous and small-scale decision making about what counts as “compliant.”

They’ve Got the Range
No matter how broad, three categories can’t capture the full range of our researchers’ output. Below is a full list of their collective output over the last year (the names of D&S-affiliated authors appear in bold):

“Algorithm Auditing: The Broken Bus on the Road to Algorithm Accountability” in Conference on Secure and Trustworthy Machine Learning — Abeba Birhane, Briana Vecchione, Deb Raji, Ryan Steed, Victor Ojewale

“Auditing Work: Exploring the New York City Algorithmic Bias Audit Regime” in FAccT ‘24 — Lara Groves, Jacob Metcalf, Alayna Kennedy, Briana Vecchione, Andrew Strait

“The Backend Work of Data Subjects: Ordinary Challenges of Living with Data in India and the US” in Media Backends: Digital Infrastructures and Sociotechnical Relations (University of Illinois Press) — Ranjit Singh

“Building Blockchain Frontiers: Ethereum as an Extension of the Californian Ideology” in International Journal of Communication — Ann Brody, Tamara Kneese, Julie Frizzo-Barker

“Carbon Emissions in the Tailpipe of Generative AI” in Harvard Data Science Review — Tamara Kneese, Meg Young

“A Collective Work Agenda for the Digital Economy” for Friedrich-Ebert-Stiftung Foundation — Aiha Nguyen

“Dead Links” in Public Books — Tamara Kneese

“Good Technology is Slow (to Scale)” in The Good Robot: Why Technology Needs Feminism (Bloomsbury) — Ranjit Singh

“Intermediaries as Infrastructure: Interrogating the Phatic Labor of State-Building” in Journal of Sociology — Ranjit Singh

“Introduction for the special issue ‘Ideologies of AI and the Consolidation of Power’: Naming Power” in First Monday — Jenna Burrell, Jacob Metcalf

“The Long Shadow of Workplace Surveillance” in Stanford Social Innovation Review — Aiha Nguyen, Wilneida Negrón

“Null Compliance: NYC Local Law 144 and the Challenges of Algorithm Accountability” in FAccT ‘24 — Lucas Wright, Roxana Muenster, Briana Vecchione, Tianyao Qu, Pika (Senhuang) CAI, Alan Smith, COMM/INFO 2450 Student Investigators, Jacob Metcalf, J. Nathan Matias

“Of Techno-Ethics and Techno-Affects” in Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines (Oxford University Press)— Sareeta Amrute

“Ordinary Ethics of Governing AI” for Carnegie Endowment for International Peace — Ranjit Singh

“Our Futures Are Interwoven” in The AI Anarchies Book (Holo) — Mimi Ọnụọha, Tiara Roxanne

“Participation in the Age of Foundation Models” in FAccT ‘24 — Harini Suresh, Emily Tseng, Meg Young, Mary Gray, Emma Pierson, Karen Levy

“Participation Versus Scale: Tensions in the Practical Demands on Participatory AI” in First Monday — Meg Young, Upol Ehsan, Ranjit Singh, Emnet Tafesse, Michele Gilman, Christina Harrington, Jacob Metcalf

“The Politics of Seamlessness: A Rights Claims Perspective on Digital Identification Technologies” in Digitalization in Practice: Intersections, Implications and Interventions (De Gruyter) — Baki Cakici, Alena Thiel, Ranjit Singh

Scaling Up Mischief: Red-Teaming AI and Distributing Governance” in Harvard Data Science Review — Jacob Metcalf, Ranjit Singh

“Seattle, Dispossession, Sensing Hate, 2016” in Naked Footnotes: A Rough Guide to Ethnographic Writing (University of Minnesota Press) — Sareeta Amrute

“Subtle Misogyny Detection and Mitigation: An Expert-Annotated Dataset” in Findings of the Association for Computational Linguistics ACL 2024– Brooklyn Sheppard, Anna Richter, Allison Cohen, Elizabeth Allyn Smith, Tamara Kneese, Carolyne Pelletier, Ioana Baldini, Yue Dong

“Territorial Dehiscence (Tearing Apart the Archive)” in Pia Arke (2024 exhibition catalog) — Tiara Roxanne

“Towards AI Accountability Infrastructure: Gaps and Opportunities in AI Audit Tooling” — Victor Ojewale, Ryan Steed, Briana Vecchione, Deb Raji, Abeba Birhane

“Working Against the Clock: Digital Surveillance in US Medicaid Homecare Services” in Journal of Sociology — Alexandra Mateescu