Mnemonic evaluative frameworks in scholarly publications: A cited reference analysis across disciplines and AI-mediated contexts

Mnemonic evaluative frameworks have become central to information literacy instruction for assessing information credibility. Widely used tools such as CRAAP, CARS, ACT UP, and SIFT remain underrepresented in scholarly literature and insufficiently aligned with emerging information challenges. This study uses cited reference analysis in the Scopus database to examine 16 mnemonic evaluative frameworks across 280 peer-reviewed journal articles, conference papers, and review articles. Citation patterns were analyzed by year, discipline, institutional affiliation, and source title. Findings reveal that while legacy models like CRAAP and CARS retain the most citations, newer frameworks such as SIFT and RADAR are proportionally more cited in AI-related literature. A subset of 49 AI-focused citing documents indicates a disciplinary shift from Library and Information Sciences toward Computer Science, Engineering, Business, and Decision Sciences since 2022. These results highlight the need for adaptive, systems-aware models that address credibility challenges associated with generative AI and algorithmic curation. In response, this study introduces the CAT Test (Check, Ask, Think), a three-part evaluative framework designed to help learners assess AI-generated content by corroborating claims, interrogating model reasoning, and reflecting on platform influence. The findings inform instructional design and contribute to ongoing conversations about algorithmic transparency and credibility in academic librarianship.