Evidence types differ systematically in probative value: detection outputs cluster at weak inferential force while admissions, observed exam behaviour and fabricated references carry the strongest profiles¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Detection tool outputs produced uniformly weak inferential force profiles, whereas interview admissions (97.2% Strong), observed prohibited exam behaviour and fabricated references carried the strongest inferential force. → Albert Munoz 2026 q2 i? Interview inconsistencies showed near-uniform ratings (Moderate relevance, High credibility, Moderate inferential force) across 98.5% of items. → Albert Munoz 2026 (2)

Evidence¶

Albert Munoz 2026¶

Albert Munoz, Mercedez Hinchcliff, Cameron Langfield and Ann Rogerson. (2026). How strong is the evidence in generative AI-related academic misconduct allegations? A mixed-methods analysis. International Journal for Educational Integrity. https://doi.org/10.1007/s40979-026-00235-9

q2 · i? · associational · r2

Credential ratings across 1,855 coded items in 1,162 cases showed Turnitin/Similarity Reports at 100% Weak inferential force and AI-detection outputs at 92.5% Weak; the strongest types "do not depend on probabilistic classification of text characteristics".

"Detection tool outputs, comprising text-similarity reports (e.g., Turnitin) and standalone AI-detection tools (e.g., GPTZero), were present but produced uniformly weak inferential force profiles, consistent with their documented limitations (e.g., Weber-Wulff et al. 2023)."

Albert Munoz 2026 (2)¶

Albert Munoz, Mercedez Hinchcliff, Cameron Langfield and Ann Rogerson. (2026). How strong is the evidence in generative AI-related academic misconduct allegations? A mixed-methods analysis. International Journal for Educational Integrity. https://doi.org/10.1007/s40979-026-00235-9

q2 · i? · associational · r2

The credential analysis found interview-inconsistency items uniformly rated because each documents inability to explain one's own work, which is credible but "equally consistent with alternative explanations such as poor comprehension or test anxiety".

"evidence related to interview inconsistencies exhibits near-uniform probative quality ratings (Moderate relevance, High credibility, Moderate inferential force) across 98.5% of items, reflecting the narrow epistemic range of this evidence type."

Discussion¶

Related Claims¶

• AI-detection tool outputs as submitted evidence collapsed to 0.5% of items in 2025 after peaking at 8.8% in 2024 — related
• Co-occurrence of fabricated references with interview admissions grew from 3.1% to 15% of cases, reaching about 1 in 7 cases by 2025 — related
• AI output errors are frequent: 44.5% fabricated references in drafted research proposals and 5–13% unsafe chatbot medical answers — related
• GenAI-related misconduct allegations rose from 11.9% of all misconduct cases in 2023 to 36.9% in 2025 against a stable total caseload — related
• Simple paraphrasing and self-editing counter-measures collapse AI detection rates to near-random levels — related
• Review-attributed evidence: fabricated model-generated references clustered in disciplinary journals in a student verification exercise — related
• AI-generated ethics exercise materials carried external citations that did not survive checking, ranging from misattributed to apparently fabricated — related
• Different evidence types differ in how strongly they can support claims about effectiveness — related
• Popular plagiarism detection tools detect plagiarism in AI-assisted essays in less than 15% of cases — related
• Assessment validity is a deeper problem than AI-output detection, and disallowed use clusters with time pressure and unclear expectations — related