Abstract Despite the proliferation of computer-generated texts, the concept of AI authorship remains ambiguous. This paper investigates evolving public perceptions of authorship pertaining to computer-generated texts through review of an international survey distributed in 2024/2025. Comparing results of this survey to those from the same survey distributed in 2017/2018, this paper offers what we believe to be the only longitudinal empirical consideration of AI authorship pre- and post-LLM boom. Even with dramatic technological advancements in natural language generation, this study suggests that societal understandings of authorship have only seen minor change since the survey’s first distribution. Participants continue to hold ambivalent opinions of AI authorship, perhaps partly due to the cultural and historical weight of the term ‘authorship’ itself. This ambivalence raises broader questions about the persistent usage of ‘authorship’ as a default term for describing computational text production. Through an interdisciplinary analysis represented by humanities and computational researchers, merging both quantitative and qualitative data, this paper interrogates whether the concept of authorship remains viable in discussions of computer-generated texts via feedback from survey participants.
Henrickson et al. (Tue,) studied this question.