Problembasiertes Lernen (PBL) und projektbasiertes Lernen (PjBL) werden seit Jahrzehnten praktiziert. Beide Ansätze betonen die aktive Auseinandersetzung mit offenen Problemen als zentralen Motor für tiefes Verständnis (Savery & Duffy, 1995). Mit der zunehmenden Einbettung von Bildungstechnologie und KI in Lernumgebungen drängt sich eine Frage auf: Können diese Werkzeuge tatsächlich erweitern, was PBL und PjBL möglich machen?
Trotz ihres gemeinsamen konstruktivistischen Fundaments unterscheiden sich beide Ansätze deutlich. Während PBL den Lernprozess in den Mittelpunkt stellt — Hypothesenbildung, selbstgesteuerte Recherche und strukturierte Reflexion —, führt PjBL diese Logik bis zu einem greifbaren Ergebnis weiter und verlangt von Lernenden ein Artefakt, das ihr Verständnis öffentlich sichtbar macht (Krajcik & Shin, 2014). Beide Ansätze sind durch jahrzehntelange Evidenz gestützt, die Zugewinne in der Anwendung von Wissen, im kritischen Denken und in der kooperativen Problemlösung belegt (Hmelo-Silver, 2004; Kokotsaki et al., 2016).
Doch beide teilen eine hartnäckige Umsetzungshürde: den Engpass der Lernbegleitung. Wirksames PBL hängt von einer erfahrenen Begleitperson ab, die metakognitives Denken modelliert, Wissenslücken sichtbar macht und Inquiry stützt, ohne das Denken der Lernenden zu ersetzen — eine ressourcenintensive Rolle, die sich in der Breite kaum aufrechterhalten lässt (Hmelo-Silver, 2004).
Genau hier könnten KI-Tutoren strukturell etwas beitragen. Statt Lehrende zu ersetzen, könnte ein gut gestaltetes KI-System mehrere Funktionen der Lernbegleitung kontinuierlich und skalierbar übernehmen: Lernende dazu anregen, ihre Lernfragen zu formulieren, bevor Inhalte geliefert werden; sokratische Rückfragen stellen, die Denkprozesse nach außen tragen; und formatives Feedback geben, das auf die aktuelle Hypothese der Lernenden abgestimmt ist. Wichtig dabei: KI könnte auch das reflexive Scaffolding tragen, das Gruppen ohne dedizierte Begleitperson typischerweise aufgeben — strukturierte Impulse, die Erfahrung in übertragbares Wissen überführen.
Das Ergebnis wäre eine Erweiterung der Reichweite von Lehrenden, kein Ersatz: KI würde die iterative, metakognitive Schicht der Begleitung übernehmen, während Lehrende sich auf Problemdesign, Community-Bildung und höhere Mentorenrollen konzentrieren, die kein Algorithmus replizieren kann. Zusammen halten menschliche und KI-gestützte Lernbegleitung ein echtes Versprechen bereit: PBL und PjBL zugänglicher, skalierbarer und nachhaltiger wirksam zu machen.
Literatur
Hmelo-Silver, C. E. (2004). Problem-based learning: What and how do students learn? Educational Psychology Review, 16(3), 235–266. https://doi.org/10.1023/B:EDPR.0000034022.16470.f3
Kokotsaki, D., Menzies, V., & Wiggins, A. (2016). Project-based learning: A review of the literature. Improving Schools, 19(3), 267–277. https://doi.org/10.1177/1365480216659733
Krajcik, J. S., & Shin, N. (2014). Project-based learning. In R. K. Sawyer (Ed.), The Cambridge handbook of the learning sciences (2nd ed., pp. 275–297). Cambridge University Press. https://doi.org/10.1017/CBO9781139519526.018
Savery, J. R., & Duffy, T. M. (1995). Problem based learning: An instructional model and its constructivist framework. Educational Technology, 35(5), 31–38.
Problem-based learning (PBL) and project-based learning (PjBL) have been practiced for decades, both emphasising active engagement with open-ended problems as the primary driver of deep understanding (Savery & Duffy, 1995). As educational technology and AI become increasingly embedded in learning environments, a pertinent question follows: can these tools meaningfully extend what PBL and PjBL make possible?
Despite their shared constructivist foundation, the two approaches are meaningfully distinct. Where PBL centres the learning process — guiding students through hypothesis generation, self-directed inquiry, and structured reflection — PjBL extends this logic toward a tangible outcome, requiring learners to produce an artifact that publicly demonstrates their understanding (Krajcik & Shin, 2014). Both are supported by decades of evidence showing gains in knowledge application, critical thinking, and collaborative problem-solving (Hmelo-Silver, 2004; Kokotsaki et al., 2016).
Yet both share a persistent implementation challenge: the facilitator bottleneck. Effective PBL depends on a skilled facilitator who can model metacognitive reasoning, surface learning gaps, and scaffold inquiry without substituting for the learner's thinking — a resource-intensive role that is difficult to sustain at scale (Hmelo-Silver, 2004).
This is precisely where AI tutors could offer a structurally meaningful contribution. Rather than replacing teachers, a well-designed AI system could perform several facilitation functions continuously and at scale: prompting learners to articulate their learning issues before content is delivered, asking Socratic follow-up questions that externalise reasoning, and providing formative feedback calibrated to the learner's current hypothesis. Crucially, AI could also sustain the reflective scaffolding that groups typically abandon without a dedicated facilitator — structured prompts that convert experience into transferable knowledge.
The result would be an extension of the teacher's reach, not a replacement: AI handling the iterative, metacognitive layer of facilitation while educators focus on problem design, community building, and the higher-order mentorship that no algorithm can replicate. Together, human and AI facilitation hold genuine promise for making PBL and PjBL more accessible, more scalable, and more consistently meaningful for learners.
References
Hmelo-Silver, C. E. (2004). Problem-based learning: What and how do students learn? Educational Psychology Review, 16(3), 235–266. https://doi.org/10.1023/B:EDPR.0000034022.16470.f3
Kokotsaki, D., Menzies, V., & Wiggins, A. (2016). Project-based learning: A review of the literature. Improving Schools, 19(3), 267–277. https://doi.org/10.1177/1365480216659733
Krajcik, J. S., & Shin, N. (2014). Project-based learning. In R. K. Sawyer (Ed.), The Cambridge handbook of the learning sciences (2nd ed., pp. 275–297). Cambridge University Press. https://doi.org/10.1017/CBO9781139519526.018
Savery, J. R., & Duffy, T. M. (1995). Problem based learning: An instructional model and its constructivist framework. Educational Technology, 35(5), 31–38.