Abstract:As generative AI diffuses through academia, policy-practice divergence becomes consequential, creating demand for auditable indicators of alignment. This study prototypes a ten-item, indirect-elicitation instrument embedded in a structured interpretive framework to surface voids between institutional rules and practitioner AI use. The framework extracts empirical and epistemic signals from academics, yielding three filtered indicators of such voids: (1) AI-integrated assessment capacity (proxy) - within a three-signal screen (AI skill, perceived teaching benefit, detection confidence), the share who would fully allow AI in exams; (2) sector-level necessity (proxy…
Abstract:As generative AI diffuses through academia, policy-practice divergence becomes consequential, creating demand for auditable indicators of alignment. This study prototypes a ten-item, indirect-elicitation instrument embedded in a structured interpretive framework to surface voids between institutional rules and practitioner AI use. The framework extracts empirical and epistemic signals from academics, yielding three filtered indicators of such voids: (1) AI-integrated assessment capacity (proxy) - within a three-signal screen (AI skill, perceived teaching benefit, detection confidence), the share who would fully allow AI in exams; (2) sector-level necessity (proxy) - among high output control users who still credit AI with high contribution, the proportion who judge AI capable of challenging established disciplines; and (3) ontological stance - among respondents who judge AI different in kind from prior tools, report practice change, and pass a metacognition gate, the split between material and immaterial views as an ontological map aligning procurement claims with evidence classes.
| Comments: | 14 pages, 2 tables, 1 figure |
| Subjects: | Computers and Society (cs.CY); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2511.02875 [cs.CY] |
| (or arXiv:2511.02875v1 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2511.02875 arXiv-issued DOI via DataCite |
Submission history
From: R. Yamamoto Ravenor [view email] [v1] Tue, 4 Nov 2025 06:24:47 UTC (17 KB)