Can ChatGPT be an author? Generative AI creative writing assistance and perceptions of authorship, creatorship, responsibility, and disclosure

AI and Society (forthcoming)
  Copy   BIBTEX

Abstract

The increasing use of Generative AI raises many ethical, philosophical, and legal issues. A key issue here is uncertainties about how different degrees of Generative AI assistance in the production of text impacts assessments of the human authorship of that text. To explore this issue, we developed an experimental mixed methods survey study (N = 602) asking participants to reflect on a scenario of a human author receiving assistance to write a short novel as part of a 3 (high, medium, or low degrees of assistance) X 2 (human or AI assistant) factorial design. We found that, for a human author, the degree of assistance they receive matters for our assessments of their level of authorship, creatorship, and responsibility, but not what or who rendered that assistance, although it was more important to disclose human rather than AI assistance. However, in our assessments of the assisting agent, human assistants were viewed as warranting higher rates of authorship, creatorship, and responsibility, compared to AI assistants rendering the same level of support. These results help us to better understand emerging norms around collaborative human-AI generated text, with implications for other types of collaborative content creation.

Author's Profile

Paul Formosa
Macquarie University

Analytics

Added to PP
2024-09-30

Downloads
44 (#97,057)

6 months
44 (#94,757)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?