New Perspectives in Science Education
AI-Based STEM Education: Generation of Individualised Mathematics Tasks
Markus Lange-Hegermann, OWL University of Applied Sciences and Arts (Germany)
Tobias Schmohl, OWL University of Applied Sciences and Arts (Germany)
Alice Watanabe, OWL University of Applied Sciences and Arts (Germany)
Stefan Heiss, OWL University of Applied Sciences and Arts (Germany)
Jessica Rubart, OWL University of Applied Sciences and Arts (Germany)
Abstract
The tendency towards self-directed learning at universities is being reinforced by the COVID-19 pandemic. However, especially in the introductory phase of studies, some students show considerable deficits with regard to self-learning competences. For this reason, attempts are made in mathematics teaching from the first week of studies to prevent knowledge gaps by feedback in small groups or by correcting homework. The nevertheless high number of dropouts in STEM subjects contrasts with the growing demand in the economy for qualified graduates. We propose to address this drop-out rate by means of an instructional design based on AI algorithms which create mathematical tasks with a tailored, individual degree of difficulty for students. Our hypothesis is, that this intervention will counteract self-assessed feelings of overstraining and increases the individual motivation to study. The exercises depend on many parameters that determine the degree of difficulty. These are adjusted iteratively, based on final or intermediate results of previously processed tasks and Learning Analytics data through Bayesian optimisation.
Keywords: AI-supported task creation, STEM education, personal learning environment, Bayes optimization (BO).
References:
- McKenney, S., & Reeves, T. C. (2018). Conducting educational design research. Routledge.
- Design-Based Research Collective. (2003). Design-based research: An emerging paradigm for educational inquiry. Educational Researcher, 32(1), 5-8.
- Paassen, B., Göpfert, C., & Hammer, B. (2016). Gaussian process prediction for time series of structured data. In Proceedings of the ESANN, 24th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning.
- Kapoor, A., Burleson, W., & Picard, R. W. (2007). Automatic prediction of frustration. Interna-tional journal of human-computer studies, 65(8), 724-736.
- Schmohl, T. (2019). Selbstgesteuertes Lernen. Explorative hochschuldidaktische Formate mit Modellcharakter für vier akademische Statusgruppen. In Selbstorganisiertes Lernen an Hochschulen. Strategien, Formate und Methoden (pp. 19-40).
- Williams, C. K., & Rasmussen, C. E. (2006). Gaussian processes for machine learning (Vol. 2, No. 3, p. 4). Cambridge, MA: MIT press.
