Since the broad public launching of artificial intelligence (AI)-based large language models in autumn 2022, a debate about potential benefits and risks of AI in education, including vocational education and training (VET) arose (cp. Chiu et al. 2023, Nemorin et al. 2023, Windelband 2023). But, as there is only little experience and almost no evidence referring to this technology in education, most publications discuss potential developments and are based on estimations. A broad consensus is, that AI will have serious influence on teaching, training and learning – but if this influence appears as threat or potential often depends strongly on the beliefs of the authors. Additionally, the various dimensions of complex teaching and learning processes might be tackled very different by AI.

Against this background, a transnational consortium with colleagues from Spain, Portugal, Slovenia and Germany decided to deliver a small piece of evidence about the usefulness of AI in a very concrete setting:

Can AI support drop-out prevention in electronic learning (e-learning) via personalised tutoring?

Drop-out rates in e-learnings are high, cp. for example Khali & Ebner (2014) or Dopler et al. (2023). Among the various potential reasons for drop-out is one, that can be influenced by (human or artificial) tutors: If the learner is lost at a certain point, individual support might guide him or her back on the track. 

To work on the question, we have chosen various e-learnings, one focussing on additive manufacturing (AM) that has been developed in a previous project (metals 2019). Target groups are apprentices in technical domains, their participation is voluntarily and completely anonymous (low-stakes), they log-in on devices of their VET-centres with functional e-mails (“user 1”). They are free to choose of 27 modules – they can work on any amount of the modules and can start where they want to start. Each module takes approx. one hour and can be completed via a short multiple-choice test. Navigation within the modules is also up to the learners; there is a suggested sequence, but it is not mandatory to follow the suggestion. Finally, each module offers additional optional materials; for example, links to explanatory videos. Or, to put it different, whilst designing the e-learning modules a high degree of freedom for the learners has been installed. All navigation patterns of the learners are tracked via the internal tracking function of the learning management system (LMS).

First pilots with two German classes of industrial mechanics are very promising. The participants represent a broad spectre from being not interested in AM (and thus not in the modules), via pragmatic and efficient work on the modules till engaged learning with many modules and the additional optional materials. Data has been analysed traditionally (comparison of navigation and correlation of patterns, without AI) and some indicators for success respective drop-out have been identified, for example that learning with certain of the offered optional materials increase the success rate in the tests – thus a traditional approach towards individualised tutoring could be to recommend these optional materials to apprentices who struggle with the test. Currently the AI is fed with the collected data, we hope that it will identify more complex navigation patterns that lead to success respective drop-out – and that analysis of these patterns will lead to more elaborated approaches of individualised tutoring.

Chiu, T. K., Xia, Q., Zhou, X., Chai, C. S., & Cheng, M. (2023). Systematic literature review on opportunities, challenges, and future research recommendations of artificial intelligence in education. Computers and Education: Artificial Intelligence, 4, 100118. Dopler, S., Beil, D., & Putz-Egger, L. M. (2023). Cognitive learning outcomes of virtual vs. in-person gamified workshops: A pre-post survey experiment. Khalil, Hanan & Ebner, Martin. (2014). MOOCs Completion Rates and Possible Methods to Improve Retention - A Literature Review. Metals (2019). https://metals.mobil-lernen.com/de/elearning Nemorin, S., Vlachidis, A., Ayerakwa, H. M., & Andriotis, P. (2023). AI hyped? A horizon scan of discourse on artificial intelligence in education (AIED) and development. Learning, Media and Technology, 48(1), 38-51. Windelband, L. (2023). Artificial Intelligence and Assistance Systems for Technical Vocational Education and Training–Opportunities and Risks. In New Digital Work: Digital Sovereignty at the Workplace (pp. 195-213). Cham: Springer International Publishing.