LLM-generated responses were substantially longer than authentic student submissions, suggesting response length as a signal the detector may exploit¶
Claim · All claims
Evidence · 1 study · causalr1·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¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Across both textual exercises and the programming exercises, LLM-generated responses were substantially longer than real student responses, suggesting length may be one signal the detector can exploit. → Juho Leinonen 2026
Evidence¶
Juho Leinonen 2026¶
Juho Leinonen, Paul Denny. (2026). Distinguishing Artificial from Authentic: Evaluating LLMs for Detecting LLM-Generated Content. https://arxiv.org/abs/2607.20446 doi:10.48550/arxiv.2607.20446
q2 · i? · causal · r1
Descriptive length comparison of real versus GPT-4o-generated submissions across the four exercises. Reflection LLM responses were roughly 3.5–5× longer (1577.9–2329.4 vs 439.7 characters) and short answers about 2–3× longer (203.7–305.7 vs 91.6 characters). No effect size is printed for this descriptive comparison.
"For reflection, LLM responses were roughly 3.5–5×longer on average (1577.9–2329.4 vs. 439.7 characters). For short answers, LLM responses were about 2–3×longer on average (203.7–305.7 vs. 91.6)."
Discussion¶
Related Claims¶
- LLM self-detection performance is strongly task-dependent: reliable for programming tasks and longer reflective responses but poor for short-answer questions — related
- For short-answer questions, real student responses received higher LLM-likelihood estimates than LLM-generated responses, a reversal of the pattern in other tasks — reports the opposite
- EchoPrompt's AUROC rises with text length, from an average of 75.8% in the shortest 1-40 word bin to 98.9% in the 321-360 bin on DetectRL Length — related
- LLM-generated programming code was reliably detected, but inserting beginner bugs had an inconsistent effect across the two programming questions — related
- Baseline LLM reflective responses were highly detectable, and persona and typo prompts progressively reduced separability from real student responses — related
- A graded likelihood detector prompt produced more consistent separation than a binary yes/no detector prompt — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
llm-response-length-detection-cue |
title |
LLM-generated responses were substantially longer than authentic student submissions, suggesting response length as a signal the detector may exploit |
description |
LLM-generated responses were substantially longer than authentic student submissions, suggesting response length as a signal the detector may exploit |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-11 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — juho-leinonen-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.