Tool: AlCoFe

AICoFe (an acronym for AI-based Collaborative Feedback system) is a system developed to improve peer and self-assessment processes in educational contexts, particularly focusing on the development of technical and soft skills in university students. It integrates customized rubrics, video recordings, Learning Analytics dashboards, and personalized feedback generated by GePeTo to create a more comprehensive and reflective assessment experience.
The system allows both professors and students to evaluate using rubrics that combine quantitative scores with qualitative observations. These evaluations are then processed by an enhanced version of GePeTo, which leverages a fine-tuned ChatGPT model to generate personalized feedback for each student. This feedback highlights strengths, areas for improvement, and specific recommendations for future performance, helping students better understand how to improve their skills. All data are anonymized, securely stored, and used to enrich the feedback process.
AICoFe includes several dashboards: one for evaluators, which facilitates consistent and rubric-based evaluations; one for students, which supports self-assessment by allowing them to view their own video recordings and compare their self-ratings with peer and instructor feedback; and one for administrators, which enables course and rubric management, peer evaluation assignments, and manual feedback validation.
A case study aimed at improving oral presentation skills, conducted with final-year engineering students at Universidad Autónoma de Madrid, demonstrated that AICoFe delivers feedback perceived as clear, specific, and actionable. Students highlighted the value of the insights for improving future presentations and for preparing their final degree project defenses.
In summary, AICoFe is a robust and flexible educational tool that modernizes feedback by combining human and AI-based input, promoting reflective learning, skill development, and increased engagement through actionable insights supported by learning analytics.

Authors: Ruth Cobos, Álvaro Becerra