Abstract:

Traditional assessment models in higher education, particularly in foundational STEM subjects like Calculus, often rely on summative examinations that measure knowledge at a single point in time. This approach provides limited feedback for student improvement, generates anxiety, and places a significant grading burden on instructors, forcing them into the role of evaluators rather than mentors. This paper presents a transformative framework that moves from a static "evaluation map" to a dynamic, AI-powered "Intelligent Learning Ecosystem" designed to foster continuous formative assessment.

The proposed ecosystem was designed for a second-semester Calculus course in an undergraduate engineering program. It replaces the traditional exam-based structure with a portfolio of six interconnected, digitally-mediated assessment instruments:
1) an initial adaptive diagnostic test to identify knowledge gaps;
2) weekly adaptive practice modules providing instant, personalized feedback;
3) collaborative, real-world engineering challenges to promote authentic application;
4) AI-assisted peer assessment of these challenges to develop critical judgment;
5) a periodic reflection portfolio to foster metacognition;
6) and a final, open-book synthesis problem to assess knowledge transfer.

Artificial Intelligence (AI) acts as the catalyst for this ecosystem, not as a replacement for the teacher. AI automates the delivery of personalized practice, provides immediate, scalable feedback, and generates actionable learning analytics for the instructor. This technological integration enables a fundamental shift in the teacher's role—from a corrector of mechanical tasks to an architect of learning experiences and a mentor for personalized student accompaniment.

Our analysis shows that this model not only enhances the student learning experience by making evaluation a continuous, low-stakes, and integrated part of the learning process, but it also optimizes the instructor's workload. We estimate a 20% reduction in the total time dedicated by the professor, redirecting effort from repetitive grading towards high-value pedagogical activities like data analysis, qualitative feedback, and personalized mentoring. This paper provides a detailed blueprint for the implementation of this ecosystem, including instrument design, time-effort dimensioning for both students and faculty, and a practical roadmap for its progressive adoption. The framework offers a scalable and human-centered solution to leverage AI for a more meaningful and effective formative assessment in higher education.

Keywords:

Formative Assessment, Artificial Intelligence in Education, Learning Analytics, STEM Education, Higher Education, Pedagogical Innovation.