How to write a CHI paper (asking for a friend)

TL;DR

This paper analyzes CHI paper structure from 1997 to 2019, focusing on heading usage and writing conventions to reveal how conventions are upheld within the community. It introduces KITSUNE, an LLM-based tool trained on CHI 2023 papers to reframe informal writing into CHI-like language and to stimulate discussion about AI-assisted authorship. The authors discuss ethical, inclusivity, and policy implications of AI in scholarly writing, urging a broader examination of what counts as acceptable CHI scholarship. Key contributions include a data-driven heading taxonomy, analysis of writing conventions over time, and a provocative AI-assisted tool to spark dialogue about future publishing norms in CHI.

Abstract

Writing and genre conventions are extant to any scientific community, and CHI is no different. In this paper, we present the early phases of an AI tool we created called KITSUNE, which supports authors in placing their work into the format of a CHI paper, taking into account many conventions that are ever-present in CHI papers. We describe the development of the tool with the intent to promote discussion around how writing conventions are upheld and unquestioned by the CHI community, and how this translates to the work produced. In addition, we bring up questions surrounding how the introduction of LLMs into academic writing fundamentally change how conventions will be upheld now and in the future