Analysing AI Utilisation in Education through Learner Question Types: A Constructivist Approach
Saved in:
| Title: | Analysing AI Utilisation in Education through Learner Question Types: A Constructivist Approach |
|---|---|
| Language: | English |
| Authors: | Hyunmin Lee (ORCID |
| Source: | Australasian Journal of Educational Technology. 2026 42(1):79-96. |
| Availability: | Australasian Society for Computers in Learning in Tertiary Education. Ascilite Secretariat, P.O. Box 44, Figtree, NSW, Australia. Tel: +61-8-9367-1133; e-mail: info@ascilite.org.au; Web site: https://ajet.org.au/index.php/AJET |
| Peer Reviewed: | Y |
| Page Count: | 18 |
| Publication Date: | 2026 |
| Document Type: | Journal Articles Reports - Research |
| Education Level: | Higher Education Postsecondary Education |
| Descriptors: | Artificial Intelligence, Technology Uses in Education, Questioning Techniques, Higher Education, Classification, Models, Barriers, Constructivism (Learning), Natural Language Processing, Information Technology, Foreign Countries, Undergraduate Students, Graduate Students |
| Geographic Terms: | Australia |
| DOI: | 10.14742/ajet.10657 |
| ISSN: | 1449-3098 1449-5554 |
| Abstract: | This study investigated the evolving role of artificial intelligence (AI) in higher education by analysing learner-generated questions through a constructivist framework. Drawing on Piaget and Vygotsky's theories, student inquiries were categorised into three roles: knowledge transmitter, facilitator and co-learner. Data from 11 students across 12 information technology courses yielded 434 authentic questions, expert labelled and augmented to balance class distributions. Several natural language processing models including bidirectional encoder representations from transformers (BERT; baseline and fine-tuned), disentangled attention BERT approach (DeBERTa) and robustly optimised BERT approach (RoBERTa) were evaluated for their ability to classify these questions. Results indicate that while models excel at processing factual (knowledge transmitter) queries, they face challenges distinguishing higher-order facilitator and co-learner questions. Notably, DeBERTa achieved the highest overall accuracy (86.36%) yet struggled with capturing contextual nuances inherent in complex queries. These findings underscore the potential of AI to support personalised learning and adaptive feedback in educational settings while highlighting the indispensable role of human oversight. Implications for integrating such models into learning management systems and avenues for future research including model refinement, cross-disciplinary validation and ethical AI implementation are discussed. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Accession Number: | EJ1500104 |
| Database: | ERIC |
| Abstract: | This study investigated the evolving role of artificial intelligence (AI) in higher education by analysing learner-generated questions through a constructivist framework. Drawing on Piaget and Vygotsky's theories, student inquiries were categorised into three roles: knowledge transmitter, facilitator and co-learner. Data from 11 students across 12 information technology courses yielded 434 authentic questions, expert labelled and augmented to balance class distributions. Several natural language processing models including bidirectional encoder representations from transformers (BERT; baseline and fine-tuned), disentangled attention BERT approach (DeBERTa) and robustly optimised BERT approach (RoBERTa) were evaluated for their ability to classify these questions. Results indicate that while models excel at processing factual (knowledge transmitter) queries, they face challenges distinguishing higher-order facilitator and co-learner questions. Notably, DeBERTa achieved the highest overall accuracy (86.36%) yet struggled with capturing contextual nuances inherent in complex queries. These findings underscore the potential of AI to support personalised learning and adaptive feedback in educational settings while highlighting the indispensable role of human oversight. Implications for integrating such models into learning management systems and avenues for future research including model refinement, cross-disciplinary validation and ethical AI implementation are discussed. |
|---|---|
| ISSN: | 1449-3098 1449-5554 |
| DOI: | 10.14742/ajet.10657 |
