The growing demand for large scale labelled data, critical for AI development, has created an underclass of workers in the global peripheries, working in often exploitative conditions, to create the datasets required to develop machine learning algorithms. At the same time, policymakers and data labelling firms themselves hail the potential of data labelling to create jobs and offer pathways to development. This is particularly critical in the case of northeast India, which has an uneasy relationship with the Indian State, characterized by long periods of armed conflict and militarization. The region is marked by “backwardness” and “underdevelopment” and the Indian government often uses “development” as a criterion to grant statehood to the largely tribal populations in the region.

This paper examines if data labelling offers viable employment opportunities in the northeast Indian state of Meghalaya, if it can offer a pathway to development, and importantly, how different actors negotiate how development is defined in this context. Through interviews and participant observation with firms that offer data labelling services, the state and central government, educational institutions, NGOs, and recruitment agencies, I investigate the experiences of data workers in Meghalaya, and how narratives of development, nationality, indigeneity, and imperialism shape their experiences. I ask: What kind of opportunities does data labelling offer workers in Meghalaya? How do various institutions make these opportunities available to workers? How is “opportunity” variously defined? How do each of these actors define “development”?

This paper examines the complex dynamics of online transgender naming practices among individuals in China, focusing on how naming functions as a technology of the self in constructing and negotiating gender identity. Drawing on 75 qualitative interviews and participatory observation, the study identifies three primary types of naming practices: online-initiated, offline-initiated, and plural naming. These practices reflect a strategic balance between social recognisability, identity fluidity, and identifiability. Online-initiated names often begin as pseudonyms in digital spaces and may later extend to offline interactions, while offline-initiated names are typically more persistent and used across multiple contexts. Plural naming involves using different names in various social spaces to manage identity strategically. The research argues that online naming is not merely a performance of an idealised self but a performative negotiation of authenticity through strategic interactions with different social contexts. Additionally, naming practices highlight the importance of connection over differentiation, as they are influenced by socialisation processes and the desire to belong to specific communities. The findings suggest that transgender authenticity is constructed through negotiation between identifiable and unidentifiable representations, past and present continuity, and the interplay of real and virtual identities. This study underscores the importance of understanding transgender naming as a socialisation process that enables individuals to navigate authenticity and connect with chosen communities, rather than solely as an individual act of self-identification.

Despite growing international diversity of computer scientists, a performance of “appropriate” English remains a requirement for scholarship and career growth in this discipline. As such, diversity advocates have lauded automatic writing-assistance software, such as ChatGPT, as a rupture which will allow multilingual scholars to collaborate in a monolingual scientific publishing ecosystem. This assimilatory theory of change suggests that scientists previously separated by linguistic boundaries would use automated translation tools to collaborate throughout the scientific process.

In this study, we seek to understand how computer scientists with different language backgrounds have collaborated differently since ChatGPT became available. Do scientists from certain regions collaborate more when ChatGPT becomes available? In what ways do scientists pursue collaboration differently with the availability of this tool, even if they choose not to use it? To answer these questions, we examine the case study of International Conference on Learning Representations (ICLR), a highly ranked, peer-reviewed computer science conference attended by thousands of academic and industry researchers from around the world. We conduct semi-structured interviews with language-marginalized authors from across five continents to understand shifting collaboration techniques. Surprisingly, interviews indicate cases of the tool being adopted precisely to prevent more cross-lingual social interaction.