Terms of inclusion: Data, discourse, violence

By Anna Lauren Hoffmann. 2020

179 citations recorded by the source corpus.

Inclusion has emerged as an early cornerstone value for the emerging domain of “data ethics.” On the surface, appeals to inclusion appear to address the threat that biased data technologies making decisions or misrepresenting people in ways that reproduce longer standing patterns of oppression and violence. Far from a panacea for the threats of pervasive data collection and surveillance, however, these emerging discourses of inclusion merit critical consideration. Here, I use the lens of discursive violence to better theorize the relationship between inclusion and the violent potentials of data science and technology. In doing so, I aim to articulate the problematic and often perverse power relationships implicit in ideals of “inclusion” broadly, which—if not accompanied by dramatic upheavals in existing hierarchical power structures—too often work to diffuse the radical potential of difference and normalize otherwise oppressive structural conditions.

Subfields: Artificial Intelligence, Safety Research, Sociology and Political Science, AI Ethics, Fairness & Society, Social Sciences & Humanities

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety Research

In Sociology and Political Science

In AI Ethics, Fairness & Society

In Social Sciences & Humanities

Citations in the corpus

BigScience: A Case Study in the Social Construction of a Multilingual Large Language Model (2022, background): “Recent scholar-ship has also explored how traditional discourses of inclusion can reinforce harmful frames and paradigms (Hoffmann, 2021) and how the disproportionate role of technology companies in social impact research can hobble efforts in that space (Young et al., 2022).”
CrowdWorkSheets: Accounting for Individual and Collective Identities Underlying Crowdsourced Dataset Annotation (2022, background): “Discourses of inclusion can serve to “further rather than subvert vulnerability to what might more broadly be called ‘dataviolence’" [25].”
Confronting Power and Corporate Capture at the FAccT Conference (2022, background): “and widen the scope of perspective, stakeholder language presumes equality of power among participants to a debate or consultation [28, 35].”
Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (2022, background): “When such classification relies on phenotypes, this dimension of representational harm essentializes historically contingent identities [85] through which classification systems entrench and produce meaning about what they represent [100, 111].”
Six Human-Centered Artificial Intelligence Grand Challenges (2023, unknown): “Exclusion may also refer to bias induced due to the data used for training an AI algorithm (Hoffmann, 2021), algorithmic bias and bias due to context of use (Ferrer et al., 2021), as well as due to the lack of following genuinely human-centered approaches when designing AI enabled systems (Margetis…”
Theories of “Gender” in NLP Bias Research (2022, background): “18653/v1/W19-3824 [29] Shivang Chopra, Ramit Sawhney, Puneet Mathur, and Rajiv Ratn Shah.”
The Participatory Turn in AI Design: Theoretical Foundations and the Current State of Practice (2023, background): “, hiring, healthcare, education, content moderation), there have been increasing calls to involve members of communities impacted by AI systems in their design [7, 99, 121, 129, 136, 137, 162, 178, 179, 213, 221].”
A Framework for Deprecating Datasets: Standardizing Documentation, Identification, and Communication (2021, background): “Such research describes complex ethical, social, and political considerations with which the ML community and field at large are beginning to grapple [36, 44, 94].”
Sociotechnical Harms: Scoping a Taxonomy for Harm Reduction (2022, background): “When such classification relies on phenotypes, this dimension of representational harm essentializes historically contingent identities [75] through which classification systems entrench and produce meaning about what they represent [89, 93].”