J. Patrick McGrail

Today, large language models (LLMs) can generate human-like responses from patterns derived from vast data sets, with little effort on the part of users and usually within minutes (Mohaimenul et al., 2024). They also offer users decision-making support and can even undertake analysis of textual documents themselves (Essien et al., 2024). However, they pose a risk to users of “cognitive off-loading” (Gerlich, 2025) or undertaking what passes for critical thought instead of leaving such thinking to the user. In this paper, we describe how we adapted Richard Paul and Linda Elder’s (2006) selection of the marks of critical thinking to analyze a brief literature review and simultaneously evaluate AI-generated analysis and evaluation of the same literature review applying the exact same criteria. While we note some benefits, especially with respect to summarization, we also find deficiencies in the AI analysis and propose possible solutions for strengthening critical thinking and related skills that users need when they seek the assistance of AI in analysis of texts.

Plagiarism may distress universities in the US, but there is little agreement as to exactly what constitutes plagiarism. While there is ample research on plagiarism, there is scant literature on the content of university policies regarding it. Using a systematic sample, we qualitatively analyzed 20 Carnegie-classified universities that are “Very High in Research.” This included 15 public state universities and five high-profile private universities. We uncovered highly varied and even contradictory policies at these institutions. Notable policy variations existed for verbatim plagiarism, intentional plagiarism and unauthorized student collaboration at the studied institutions. We conclude by advising that the American Association of University Professors and other overarching academic bodies confer and come to accord on the disposition of these issues