AI can improve teacher training

11 Apr 2022

Budding teachers can benefit from AI-assisted training: A study highlights the potential of adaptive feedback.

Schüler melden sich beim Unterricht.

Teachers play a critical role in recognising learning difficulties. AI-generated feedback can help to train these skills. | © Robert-Kneschke/Adobe Stock

During their studies, pre-service teachers’ often lack the opportunity to gain sufficient practical experience. Combining simulations with artificial intelligence can be a promising way to give a more hands-on edge to the skills they learn. This is the conclusion reached by a study conducted by Professor Frank Fischer, Professor for Education and Educational Psychology at LMU Munich, and Dr. Michael Sailer. The study appears in the latest edition of the journal Learning and Instruction.

The two educational researchers worked with Iryna Gurevych, Professor of Computer Science at the Technical University of Darmstadt, to develop and train what is known as an artificial neural network that gives adaptive feedback tailored to pre-service teachers’ individual performances.

As part of an experiment, online simulations were used to train the diagnostic reasoning of pre-service teachers. The aim was for trainees to learn to recognize certain learning difficulties and provide written justification of why they suspected a given difficulty. In these written justifications composed by the pre-service teachers, the AI identified what the learners had done right and wrong and gave feedback accordingly.

“Above all, using AI and providing individualized feedback improved the diagnostic reasoning of the trainee teachers,” says Michael Sailer, who believes that AI deployment adds value especially for the advancement of complex skills.

“The adaptive feedback is similar to the individual feedback that a human lecturer would give,” Frank Fischer says. “Especially in large courses of study such as teacher education but also medical education, where you have lots of students, this is a promising way to add a great deal of value.”

The study was published in collaboration with Professor Riikka Hofmann of the University of Cambridge under the aegis of the strategic partnership between LMU and Cambridge.


Michael Sailer, Elisabeth Bauer, Riikka Hofmann, Jan Kiesewetter, Julia Glas, Iryna Gurevych, Frank Fischer: “Adaptive Feedback from Artificial Neural Networks Facilitates Pre-Service Teachers’ Diagnostic Reasoning in Simulation-based Learning”. In: Learning and Instruction 2022

What are you looking for?