TLDR: A survey of digital humanists reveals that while they acknowledge the importance of disclosing Generative AI (GenAI) use in research, actual disclosure rates are low. There are differing views on what activities require disclosure and how to do it, with most believing that institutional policies, rather than individual decisions, should establish disclosure safeguards.
Generative AI (GenAI) is becoming increasingly common in academic work, including the digital humanities. While it offers many benefits, its use also brings up important questions about academic integrity and research ethics. A recent survey study explored how digital humanists view and approach the disclosure of GenAI use in their research. The findings offer valuable insights for scholars, institutions, and funders looking to create effective disclosure policies.
The study, conducted by Rongqian Ma, Xuhan Zhang, and Adrian Wisnicki, surveyed digital humanities scholars to understand their perspectives on GenAI disclosure. The researchers found that while most digital humanists recognize the importance of being transparent about using GenAI, the actual rate of disclosure in their research remains low. This suggests a gap between acknowledging the need for transparency and putting it into practice.
One interesting finding was the differing opinions on what activities most require disclosure and the best ways to disclose. For instance, a majority of respondents (63%) felt it was necessary to disclose GenAI use, with most preferring a “brief disclosure” that includes details like the model/version used and the parts generated by AI. A smaller percentage opted for “detailed disclosure” or “minimal disclosure.”
Despite this agreement on the value of transparency, only 28% of participants who believed disclosure was necessary had actually disclosed GenAI use in their research. Many of these disclosures were voluntary and infrequent, happening only 1-2 times. This discrepancy might be due to a lack of motivation, external obligations, or even a perceived “AI disclosure penalty,” where disclosing AI use could negatively impact how the work is perceived.
The survey also revealed that disclosure priorities vary depending on the research activity. “Writing and rewriting” was the most frequently prioritized activity for disclosure across most disciplines, highlighting the central role of writing in digital humanities scholarship. However, computer scientists, a subfield within digital humanities, prioritized “brainstorming” for disclosure instead.
Crucially, the study found that most digital humanists believe that the responsibility for establishing AI disclosure safeguards should fall on institutions rather than individuals. Journal publishers and editorial boards were seen as the primary gatekeepers (87%), followed closely by academic institutions (86%). Professional societies, funding bodies, and ethical-review boards also received significant support for their role in oversight, indicating a preference for a multi-layered approach to policy enforcement within the research community.
Also Read:
- Consumer Trust and AI: Why Hiding AI Use Damages Brands More Than Exaggeration
- Navigating the Ethical Landscape of AI-Generated Art
In conclusion, this survey highlights a collective understanding among digital humanists about the importance of disclosing GenAI use, but also reveals varied opinions on how to implement these disclosures effectively. The research provides empirical guidance for developing policies that can help integrate AI responsibly into academic research. You can read the full research paper for more details at this link.
