FACET ▶ ongoing

Framework for Agent-based Classroom Enhancement for Teacher

FACET is a teacher-facing multi-agent AI framework developed jointly by ZIB, the Weizenbaum Institute, and partner schools in Berlin to support differentiated instruction in heterogeneous classrooms. The project combines psychological research on human–AI interaction with algorithmic modeling and multi-agent system design.

since Aug 2025since Aug 25
8
3

🧑‍🎓 IOL Project Members

Jana Gonnermann-Müller
Principal Investigator
gonnermann-mueller (at) zib.de
Sebastian Pokutta
Principal Investigator
pokutta (at) zib.de
Moritz Igel
igel (at) zib.de
Muriel Fischer
fischer (at) zib.de
Nicolas Leins
leins (at) zib.de
Varya Geronimus
geronimus (at) zib.de

🤝 External Project Members

Jennifer Haase
Principal Investigator
jennifer.haase (at) hu-berlin.de
Max Fackeldey
Principal Investigator
fackeldey (at) math.tu-berlin.de

🔬 Project Description

FACET addresses the persistent gap between the pedagogical ideal of differentiated instruction and its feasibility under real classroom conditions. The system coordinates four specialized agents, a learner simulation agent, diagnostic reasoning agent, material generation agent, and quality assurance agent, to produce curriculum-aligned, accessibility-compliant worksheets tailored to diverse learner profiles encompassing cognitive proficiency, motivational orientation, and learning differences such as dyslexia and ADHD.

Partner schools in Berlin were involved as co-developers throughout the design process. This collaboration now extends into longitudinal deployment across partner schools.

The project researches the responsible design and integration of AI systems in schools, contributing to an evidence-based understanding of how AI can meaningfully support teachers and students in diverse educational settings.

📝 Publications and preprints

Preprints

  1. Gonnermann-Müller, J., Haase, J., Leins, N., Kosch, T., and Pokutta, S. (2026). LLM-Based Educational Simulation: Evaluating Temporal Student Persona Stability Across ADHD Profiles. [arXiv]
    [BibTeX]
    @misc{2026_JanagonnermannmuellerEtAl_Llmstudentsimulation_2605-06307,
  archiveprefix = {arXiv},
  eprint = {2605.06307},
  arxiv = {arXiv:2605.06307},
  primaryclass = {cs.HC},
  year = {2026},
  author = {Gonnermann-Müller, Jana and Haase, Jennifer and Leins, Nicolas and Kosch, Thomas and Pokutta, Sebastian},
  title = {LLM-Based Educational Simulation: Evaluating Temporal Student Persona Stability Across ADHD Profiles},
  date = {2026-05-07}
}
  2. Gonnermann-Müller, J., Haase, J., Fackeldey, K., and Pokutta, S. (2025). FACET: Teacher-Centred LLM-Based Multi-Agent Systems-Towards Personalized Educational Worksheets. [arXiv]
    [BibTeX]
    @misc{2025_JanaHaaseFackeldeyPokutta_Facetllm_2508-11401,
  archiveprefix = {arXiv},
  eprint = {2508.11401},
  arxiv = {arXiv:2508.11401},
  primaryclass = {cs.HC},
  year = {2025},
  author = {Gonnermann-Müller, Jana and Haase, Jennifer and Fackeldey, Konstantin and Pokutta, Sebastian},
  title = {FACET: Teacher-Centred LLM-Based Multi-Agent Systems-Towards Personalized Educational Worksheets},
  date = {2025-08-15}
}

Conference proceedings

  1. Gonnermann-Müller, J., Haase, J., Leins, N., Igel, M., Fackeldey, K., and Pokutta, S. (2026, January 30). FACET: Multi-Agent AI Supporting Teachers in Scaling Differentiated Learning for Diverse Students. Proceedings of the International Joint Conference on Artificial Intelligence. [arXiv]
    [BibTeX]
    @inproceedings{2026_JanagonnermannmuellerEtAl_Facetmultiagent_2601-22788,
  year = {2026},
  booktitle = {Proceedings of the International Joint Conference on Artificial Intelligence},
  archiveprefix = {arXiv},
  eprint = {2601.22788},
  arxiv = {arXiv:2601.22788},
  primaryclass = {cs.HC},
  author = {Gonnermann-Müller, Jana and Haase, Jennifer and Leins, Nicolas and Igel, Moritz and Fackeldey, Konstantin and Pokutta, Sebastian},
  title = {FACET: Multi-Agent AI Supporting Teachers in Scaling Differentiated Learning for Diverse Students},
  date = {2026-01-30}
}