Search and Filter

Argus: Academic Integrity In The Era Of Generative AI

Date

AI Overview

Why it matters

The rapid spread of large language models allows students to generate programming assignment solutions without learning concepts, challenging academic integrity enforcement. Automated detection tools risk false accusations that carry serious consequences for students.

Study at a glance
  • Question: How prevalent is large language model code generation among students, and how does it relate to exam performance?
  • Setting and data: Purdue University undergraduate C programming course, Spring 2020 and 2022–2026, 3,893 student-semester pairs across 42,695 submissions.
  • Method: System development of an automated detection tool (Argus) combined with observational correlation and descriptive analysis.
  • AI tools: Argus, ChatGPT, Claude, Gemini
Key findings
  • In Spring 2026, 45.7% of enrolled students were flagged for suspected LLM-assisted code development following automated triage and manual review.
  • Higher heuristic detection scores were significantly negatively correlated with student performance on in-person, proctored examinations in recent semesters.
  • Students who withdrew from the course after the first midterm exhibited significantly higher average detection scores than students who completed the course.

Summary drafted with AI from the full paper and may contain errors. See the full paper for details.

The rapid proliferation of large language models (LLMs) in the context of education has introduced significant challenges in enforcement of academic integrity, especially in programming courses. We present Argus, an automated detection system for LLM-assisted student work in undergraduate C programming assignments. Argus integrates behavioral and stylistic indicators to create a holistic picture of the student's progress through an assignment and surfaces anomalies that point to potential misuse of LLM assistance. We quantify and analyze data over six years of Spring semester offerings in a large-enrollment CS2 course at Purdue University using Argus, finding that 45% of enrolled students exhibited patterns consistent with LLM-assisted code development in Spring 2026. To contextualize these results, we analyze the relationship between flagged LLM use and student performance on written, in-person proctored examinations, and find a significant negative correlation. We also explore the problem of mitigating false positives, recognizing that erroneous accusations of academic integrity carry significant consequences for students and instructors alike, particularly in the context of large enrollment courses. We argue that any automated detection system must be accompanied by a structured process for human review. We discuss the consequences for future course design, changing academic policy as these tools become more ubiquitous, and the pedagogical implications of LLM-based tools in computer science education.