AI tutoring in software engineering education
Date
2026
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Kontext. Das Erlernen von Software Technik ist ein kompliziertes Thema, das effektive Lehrmethoden erfordert, um sicherzustellen, dass Studierende die notwendigen Fähigkeiten und Kenntnisse erwerben können. Zeitnahes und effektives Feedback ist dabei entscheidend, damit Studierende optimale Lernergebnisse erzielen können.
Problem. Traditionelle Lehrmethoden sind oft unzureichend darin, Feedback zeitnah und individuell bereitzustellen, was zu suboptimalen Lernergebnissen führt. Oftmals suchen Studierende auch nicht aktiv nach Feedback, was ihren Lernprozess zusätzlich behindert.
Ziel. Ziel dieser Arbeit ist die Entwicklung eines künstliche Intelligenz (KI)-Tutors, der personalisiertes Feedback geben und sich an verschiedene Lernstile anpassen kann, um die Lernerfahrung von Studierenden im Erlernen von Software Technik zu verbessern. Zusätzlich untersuchen wir proaktive Feedback Mechanismen, um den Lernprozess der Studierenden weiter zu unterstützen. Methode. Wir führen einen Requirements-Engineering-Prozess durch, um die Anforderungen der Studierenden sowie gängiger Literatur zu erarbeiten. Darauf basierend entwerfen wir einen KI-Tutor, der nicht nur reaktives, sondern auch proaktives Feedback liefert, welches jeweils auf verschiedene Spielertypen zugeschnitten ist. Der Ansatz wird in die bestehende Lernplattform MEITREX implementiert und mittels einer experimentellen Studie mit Studierenden evaluiert.
Ergebnis. Wir haben wichtige Anforderungen bezüglich eines KI-Tutors in Erlernen von Software Technik identifiziert, insbesondere im Hinblick auf die Wahrnehmung von proaktivem Feedback durch die Studierenden. Unsere Evaluation deutet darauf hin, dass Studierende von dem personalisierten Feedback des KI-Tutors profitieren, was zu einer gesteigerten Motivation führt, während die Akzeptanz proaktiver Hinweise jedoch unter den Studierenden variiert.
Fazit. Der entwickelte KI-Tutor verbessert das Erlernen von Software Technik durch die Bereitstellung von personalisiertem Feedback, das durch verschiedene Spielertypen und das aktuelle Fähigkeitsniveau auf den einzelnen Studierenden zugeschnitten ist. Während der Ansatz der Personalisierung des reaktiven Feedbacks vielversprechend erscheint, ist jedoch weitere Forschung von Nöten, damit das proaktive Feedback des KI-Tutors optimiert werden kann.
Context. Software engineering education is a complicated topic that requires effective teaching methods to ensure students acquire the necessary skills and knowledge. Timely and effective feedback is crucial for students to achieve optimal learning outcomes. Problem. Traditional teaching methods often fall short in providing feedback in a timely and individualized manner, leading to suboptimal learning outcomes. Often, students do not actively seek feedback, which further hinders their learning process. Objective. The objective of this thesis is to develop an artificial intelligence (AI)-tutor that can provide personalized feedback and adapt to different learning styles, to enhance the learning experience for students in software engineering education. Additionally, we explore proactive feedback mechanisms to further support student learning. Method. We conduct a requirements engineering process to gather insights into the needs of students and educational theory to design an AI-tutor that provides not only reactive but also proactive feedback, both tailored to different player types. The approach is implemented into the existing learning platform MEITREX and evaluated through an experimental study with students. Result. We identified key insights on requirements for an AI-tutor in software engineering education, specifically on student perception of proactive feedback. Our evaluation indicates that students benefit from the personalized feedback provided by the AI-tutor, leading to improved motivation, but acceptance of proactive hints varies among students. Conclusion. The developed AI-tutor enhances software engineering education by providing personalized feedback, tailored to the individual student through different player types and student skill. However, while the reactive feedback approach seems promising, further research is required to optimize the AI-tutor’s proactive feedback approach.
Context. Software engineering education is a complicated topic that requires effective teaching methods to ensure students acquire the necessary skills and knowledge. Timely and effective feedback is crucial for students to achieve optimal learning outcomes. Problem. Traditional teaching methods often fall short in providing feedback in a timely and individualized manner, leading to suboptimal learning outcomes. Often, students do not actively seek feedback, which further hinders their learning process. Objective. The objective of this thesis is to develop an artificial intelligence (AI)-tutor that can provide personalized feedback and adapt to different learning styles, to enhance the learning experience for students in software engineering education. Additionally, we explore proactive feedback mechanisms to further support student learning. Method. We conduct a requirements engineering process to gather insights into the needs of students and educational theory to design an AI-tutor that provides not only reactive but also proactive feedback, both tailored to different player types. The approach is implemented into the existing learning platform MEITREX and evaluated through an experimental study with students. Result. We identified key insights on requirements for an AI-tutor in software engineering education, specifically on student perception of proactive feedback. Our evaluation indicates that students benefit from the personalized feedback provided by the AI-tutor, leading to improved motivation, but acceptance of proactive hints varies among students. Conclusion. The developed AI-tutor enhances software engineering education by providing personalized feedback, tailored to the individual student through different player types and student skill. However, while the reactive feedback approach seems promising, further research is required to optimize the AI-tutor’s proactive feedback approach.