AI as anti-social media: Changing patterns in collaboration between multilingual computer scientists

• In Event: 50462 - Section on Science, Knowledge, and Technology Roundtables
  In Refereed Roundtable (60 minute): Table 1: SKAT approaches to AI

Abstract

The computer science community operates in English, from global education and hiring, to indexed research output. 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 the increasing availability and adoption of automatic writing-assistance and translation software which allow authors to instantly translate or even fully generate written manuscripts. The widespread availability such tools has inspired hope that multilingual scholars might be able to better integrate into a monolingual scientific publishing ecosystem. This theory of change suggests that scientists previously separated by linguistic boundaries would use automated translation tools to collaborate throughout the scientific process. Though there is substantial evidence of widespread use of these tools by scientists, our ongoing mixed methods study of a large computer scientific publishing community suggests that the theory of change does not hold. Collaborations across countries appear to drop since the release of ChatGPT in late 2022, and interviews with language-marginalized scientists indicate cases of the tool being adopted precisely to prevent more cross-lingual social interaction. Moreover, though our data describe a community which has seen rapidly increased internationalization during the same period; this internationalization appears to remain linguistically homophilous, and new, multilingual entrants into this scientific community describe linguistic exclusion.

Author

• Haley Lepp, Stanford University