AICoFe: Implementation and Deployment of an AI-Based Collaborative Feedback System for Higher Education

Effective peer feedback is critical for developing reflective skills in higher education, but student-generated comments often suffer from inconsistent quality. Researchers Alvaro Becerra, Alejandra Palma, and Ruth Cobos from Universidad Autónoma de Madrid have implemented and deployed AICoFe (AI-based Collaborative Feedback), a system that bridges this gap using a human-centered AI approach. AICoFe orchestrates a multi-LLM pipeline featuring OpenAI's GPT-4.1-mini, Google's Gemini 2.5 Flash, and Meta's Llama 3.1 to synthesize quantitative rubric data and qualitative observations into coherent, actionable feedback. The modular architecture ensures each model contributes its strengths—GPT-4.1-mini for structured reasoning, Gemini for speed, and Llama for open-source flexibility.

A key innovation is the "teacher-in-the-loop" mediation workflow. Educators use specialized Learning Analytics dashboards to curate and refine AI-generated drafts before delivering them to students, preserving pedagogical control. The underlying data infrastructure employs a hybrid SQL and MongoDB strategy to ensure traceability and manage the semi-structured nature of feedback versions. This allows administrators to audit which suggestions were AI-generated, teacher-edited, or student-revised. The system was accepted for presentation at LASI Spain 2026 (Learning Analytics Summer Institute) and is published on arXiv (arXiv:2605.04740). By automating the heavy lifting of feedback generation while keeping instructors in the decision loop, AICoFe promises to scale high-quality peer feedback without sacrificing educational nuance.