Detect AI-Assisted Cheating in Assignments

Problem Context

Teachers and educators submit student assignments but struggle to detect when AI tools like ChatGPT or Claude have been used. Without clear evidence, they can’t take disciplinary action, wasting time on investigations that often lead nowhere. This undermines academic integrity and gives cheaters an unfair advantage.

Pain Points

Educators notice suspicious patterns (e.g., '\*') in submissions but can’t identify which AI tool produced them. Manual searches online fail to provide answers, leaving them without proof to confront students. The lack of actionable evidence forces them to either ignore cheating or spend hours on fruitless investigations.

Impact

Time is wasted on investigations that don’t yield results, and academic integrity is compromised. Students who cheat gain unfair advantages, while honest students face an uneven playing field. The inability to prove cheating erodes trust in the education system and discourages educators from enforcing rules.

Urgency

AI tools are becoming more accessible, making cheating easier to hide. Without a way to detect and prove AI-assisted cheating, educators risk losing control over academic integrity. The problem is urgent because it directly impacts the fairness and credibility of education.

Target Audience

K-12 teachers, university professors, academic integrity officers, and educators in online learning platforms. These users face the same challenge of detecting AI cheating without clear evidence, and they are increasingly frustrated with the lack of effective tools.

Solution Approach

AI Cheat Detector is a specialized tool that scans student submissions for unique artifacts left by AI tools. It uses a proprietary database of fingerprints (e.g., notation patterns, formatting quirks) to identify which AI tool was likely used. The tool provides educators with actionable evidence to take disciplinary action, restoring fairness in assessments.

Key Features

1. Tool Identification: Matches artifacts to a database of 50+ AI tools (e.g., '92% likely ChatGPT').
2. Educator Dashboard: Flags suspicious submissions with probabilities and exportable evidence reports.
3. Crowd-Sourced Database: Users can submit new AI artifacts to expand the tool’s detection capabilities.

User Experience

Educators upload assignments via a simple dashboard. The tool scans for AI artifacts and returns a report highlighting suspicious patterns with tool probabilities. They can export evidence for disciplinary actions or mark submissions for review. The process is fast (under 1 minute per scan) and requires no technical skills.

Differentiation

Unlike plagiarism checkers, this tool focuses on AI-specific artifacts, providing clearer evidence for disciplinary actions. The proprietary database and educator-friendly reporting set it apart from generic scanning tools. It also allows users to contribute new AI fingerprints, keeping the tool updated with emerging threats.

Scalability

The tool scales with the number of AI tools and users. Schools can purchase seats for multiple educators, and the database expands as more users submit new artifacts. Future features could include integration with LMS platforms (e.g., Canvas, Moodle) for automated scanning of submissions.

Expected Impact

Educators save time by quickly identifying AI-assisted cheating with actionable evidence. Academic integrity is restored, and cheaters are held accountable. The tool reduces the administrative burden of investigations and ensures fair assessments for all students.