Paper Number
ECIS2026-1728
Paper Type
CRP
Abstract
In AI, ground truth is an objective reference point against which an AI model is trained and evaluated. By departing from the idea that ground truth is socially constructed, this article shows how gender ground truth is established through a range of intertwined data practices. Drawing on interviews with gender-aware AI professionals and discourse analysis, the paper highlights four interpretative repertoires through which interviewees made sense of gender in AI. First, the interpretative repertoire conceptual framing focuses on the definitions used in machine learning. Collection purpose highlights how and why the data was collected, while the politics of representation focuses on who is included in the data. Finally, interpretive practices emphasise the importance of data labellers’ subjectivities. The article shows how gender ground truth is established and stabilised through various practices that make gender in AI appear as objective and neutral.
Recommended Citation
Kelan, Elisabeth, "Establishing Gender Ground Truth: The Epistemic Practices Of AI Work" (2026). ECIS 2026 Proceedings. 6.
https://aisel.aisnet.org/ecis2026/resp_AI/resp_AI/6
Establishing Gender Ground Truth: The Epistemic Practices Of AI Work
In AI, ground truth is an objective reference point against which an AI model is trained and evaluated. By departing from the idea that ground truth is socially constructed, this article shows how gender ground truth is established through a range of intertwined data practices. Drawing on interviews with gender-aware AI professionals and discourse analysis, the paper highlights four interpretative repertoires through which interviewees made sense of gender in AI. First, the interpretative repertoire conceptual framing focuses on the definitions used in machine learning. Collection purpose highlights how and why the data was collected, while the politics of representation focuses on who is included in the data. Finally, interpretive practices emphasise the importance of data labellers’ subjectivities. The article shows how gender ground truth is established and stabilised through various practices that make gender in AI appear as objective and neutral.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.