Student submissions matched the LLM-generated reference bank more often in later cohorts, with 2025 incidence 1.76–2.12 times the 2021 rate, persisting among full-pass submissions¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Student–LLM match incidence was 1.06–1.19 times its 2021 rate in 2023 and 1.76–2.12 times that rate in 2025, with every annotated incidence comparison at BH-adjusted q < .001. → Runlong Ye 2026

Evidence¶

Runlong Ye 2026¶

Runlong Ye, Jing Fan, Angela Zavaleta Bernuy, Oscar Karnalim, Paul Denny, Juho Leinonen, and Michael Liut. (2026). Correctness, Convergence, and AI-Generated Code Detection: A Longitudinal Study of Student and Large Language Model Code in Introductory Programming. https://arxiv.org/abs/2610.00863 doi:10.48550/arxiv.2610.00863

q2 · i? · associational · r2

Observational cohort comparison of 29,970 final student submissions against 90,000 LLM attempts using MOSS with starter code excluded (Figure 2, Table 4). The article treats the interval tests as model-based descriptive annotations because of dependency in the data.

"Student–LLM match incidence was 1.06–1.19 times its 2021 rate in 2023 and 1.76–2.12 times that rate in 2025 (Figure 2). Every annotated incidence comparison had BH-adjusted 𝑞<. 001, including the small 2023 shifts."

Discussion¶

Related Claims¶

• Most reported student–LLM matches involved short snippets, and raising the minimum match length rapidly reduced coverage, making it a key review-workload control — related
• Frontier LLMs produced near-perfect solutions to CS1 labs, with mean hidden-instructor-test scores of 98.93–99.99% across model–year combinations — related