AI-Driven Contribution Evaluation and Conflict Resolution: A Framework & Design for Group Workload Investigation
This paper proposes a novel AI-driven framework and tool design that evaluates individual team contributions and resolves conflicts by integrating heterogeneous data into a three-dimensional benchmark system, utilizing Large Language Models to generate transparent, interpretable advisory judgments while addressing feasibility, bias, and institutional policy constraints.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are the referee of a soccer team where everyone gets the same score at the end of the season, regardless of who actually ran the most, passed the ball, or scored the goals. This is often how university group projects work. It's fair on the surface, but it feels wrong when one person does all the work while another just watches.
This paper proposes a new "smart referee" system to help teachers solve these unfair situations without spending hours digging through emails and code files. Here is how the authors break it down:
The Problem: The "Free Rider" and the "Overachiever"
In group work, two bad things happen:
- The Free Rider: Someone who doesn't do their share (like a passenger on a bus who never pays the fare).
- The Overachiever: Someone who does everything, gets resentful, and feels like they are carrying the whole team.
Currently, when students argue about who did what, teachers have to manually read hundreds of chat messages, code files, and meeting notes to figure out the truth. It's slow, stressful, and prone to human bias. Existing tools try to help, but they mostly just ask students to rate each other (which can be biased by popularity) or count simple numbers without understanding the context.
The Solution: A "Digital Detective" Framework
The authors propose a new tool that acts like a digital detective. Instead of just counting numbers, it uses Artificial Intelligence (AI) to read the "story" of the project.
They organize the evidence into three main categories, like three different lenses on a camera:
- Contribution (The "What"): Did the student actually submit work?
- The Analogy: Imagine checking a construction site. Did the student lay bricks, write the blueprints, or just stand around? The system counts the "bricks" (lines of code, words written) and checks if they are high quality.
- Interaction (The "How"): How did they talk to the team?
- The Analogy: This is like listening to the team's walkie-talkie. Was the student rude? Did they ignore messages? Did they help clarify confusion, or did they make things more chaotic?
- Role (The "Why"): Did they follow the plan?
- The Analogy: Did the student show up to the meetings? Did they organize the files? Did they stick to the schedule, or did they leave the team scrambling at the last minute?
How the "Smart Referee" Works
The system doesn't just guess; it follows a strict recipe:
- Gathering Evidence: It collects everything: code, emails, chat logs, meeting minutes, and even peer reviews.
- The "Gini" Alarm: It uses a mathematical tool called the Gini Index (usually used to measure wealth inequality) to spot unfairness.
- Scenario A: If one person did 90% of the work, the alarm rings.
- Scenario B: If one person did almost nothing while everyone else worked hard, the alarm rings.
- The AI Judge (The "Consultant"): Here is the most important part. The AI does not give the final grade. Instead, it acts like a consultant or a detective's report.
- It takes the data, the alarms, and the context, and writes a clear, transparent summary for the human teacher.
- Example: "Student A contributed 80% of the code and was very helpful in meetings, but Student B was absent from 3 meetings and their chat messages were negative. Here is the evidence."
- Human in the Loop: The human teacher reads the AI's report and makes the final decision. This ensures the system is fair and follows the rules.
The Rules of the Game (Policy & Safety)
The authors are very careful about privacy and rules. They acknowledge that schools have strict laws (like GDPR in the UK) about student data.
- No Secret Training: The AI cannot use student work to "teach" itself unless the students agree.
- Advisory Only: The tool is designed to advise the teacher, not to replace the teacher. This keeps the human in control and avoids legal trouble regarding automated decision-making.
- Bias Checks: The system tries to correct for common human biases, like students rating their friends too highly or rating enemies too lowly.
What the Paper Claims (and Doesn't Claim)
- It Claims: They have designed a framework that organizes messy data into clear categories. They believe this is technically possible and legally safe if done correctly (with consent and human oversight). They argue this would save teachers time and make conflict resolution more objective.
- It Does NOT Claim: They have not built the final, perfect product yet. They haven't tested it on thousands of students to prove it works perfectly in every real-world scenario. They are proposing the blueprint and the logic, not a finished commercial app.
In a Nutshell
This paper suggests building a tool that acts as a fairness assistant. It gathers all the messy evidence of a group project, uses AI to spot the "unfair players" based on what they did, how they talked, and how they behaved, and then hands a clear, evidence-backed report to the human teacher. The teacher then uses this report to make the final, fair decision, ensuring no one is unfairly punished or rewarded.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.