intelligent-tutoring-systems
Filtering by topic intelligent-tutoring-systems(4)Clear all filters
- PaperOpenAlex — TESOL research22 Jul 2026
ENHANCING FOREIGN LANGUAGE TEACHING METHODOLOGY THROUGH ARTIFICIAL INTELLIGENCE: A CASE STUDY OF ENGLISH LANGUAGE LEARNING
Farmonova Naimakhon Furkat kizi
A mixed-method study evaluated AI-based tools—intelligent tutoring systems, natural language processing applications, and adaptive learning platforms—for English language teaching in higher education. Results showed significant improvements in vocabulary acquisition, communicative competence, and learner autonomy, though challenges in teacher readiness, infrastructure, and ethics remain.
Original abstract
Artificial Intelligence (AI) is rapidly transforming educational practices, particularly in the field of foreign language teaching. This study explores how AI can enhance English language teaching methodology in higher education. Using a mixed-method research design, the study evaluates the impact of AI-based tools such as intelligent tutoring systems, natural language processing applications, and adaptive learning platforms on students’ language proficiency. The findings reveal that AI significantly improves vocabulary acquisition, communicative competence, and learner autonomy. However, challenges related to teacher readiness, infrastructure, and ethical considerations remain. The study contributes to the development of modern AI-based pedagogical frameworks.
- PaperEdArXiv (OSF Preprints)20 Jul 2026
Student Learning Rates When Practicing with an Intelligent Tutoring System Versus On Paper
Conrad Borchers, Qianru Lyu, Ritesh Kanchi, Kenneth R. Koedinger et al.
This study directly compared learning rates, measured as improvement per problem-solving step, between practice with an intelligent tutoring system (ITS) and paper-based practice among 97 middle-school students. ITS-supported practice yielded significantly higher overall learning gains and positive learning rates, while paper-based learning rates were near zero, especially for lower-proficiency students. The findings underscore the critical role of immediate feedback and adaptive support in enabling learning during practice.
Original abstract
Intelligent tutoring systems (ITSs) are widely used in K-12 education and are known to improve learning outcomes compared to traditional instruction. However, prior work has focused primarily on end-of-test performance, with little attention to learning processes, such as how much students learn at each practice opportunity. In particular, learning rates have been extensively studied in ITS environments but not for paper-based problem solving. We conducted a within-subjects classroom experiment with 97 middle-school students solving matched mathematics problems either with an ITS or on paper. We compared both overall learning gains and process-level learning rates, defined as improvement per problem-solving step. To enable this comparison, we introduce a novel method for converting paper-based work into step-level transaction data compatible with learning-curve modeling. Results show that ITS-supported practice led to significantly higher learning gains and substantially higher learning rates than paper-based practice. While students in the ITS condition exhibited consistent positive learning rates, learning rates in the paper condition were indistinguishable from zero, indicating little measurable learning during problem solving without feedback. These differences were especially pronounced for students with lower prior proficiency. This study provides the first direct comparison of learning rates between ITS and paper-based practice and introduces a generalizable methodology for analyzing learning processes in non-digital environments. The findings highlight the critical role of immediate feedback and adaptive support in enabling learning during practice and support the broader adoption of ITS in K-12 education.
- PaperComputers and Education: Artificial Intelligence17 Jun 2026
Artificial intelligence in vocational education and training: A systematic review of educational purposes, theoretical conceptualizations, and empirical effectiveness
Viola Deutscher, Herbert Thomann, Olga Zlatkin‐Troitschanskaia, Ulrike Weyland et al.
A systematic review of 26 empirical studies found that AI interventions like intelligent extended reality and tutoring systems improve procedural competence and motivation in vocational education, but evidence is limited by a lack of randomized experiments and overreliance on behaviorist designs. The authors argue that current research portrays a generalized success narrative, neglecting failure cases and learner agency, and call for future work to design AI that augments human judgment rather than replicating instruction.
Original abstract
Background Artificial intelligence (AI) increasingly shapes vocational education and training (VET). Despite its transformative potential, a systematic overview of the design characteristics and empirical effectiveness of AI-based interventions in vocational contexts remains lacking. Aim This review synthesizes empirical research on AI-supported interventions in VET, focusing on educational purposes, theoretical conceptualizations of human-AI interaction, methodological designs, and evidence of learning outcomes. Method Following PRISMA guidelines, 26 empirical studies published between 2015 and 2026 were identified through ERIC, Web of Science, and Elicit, and analyzed using a theory-informed coding scheme. Findings Intelligent Extended Reality shows consistent positive effects on procedural competence, practical skills, and learner motivation; Intelligent Tutoring Systems foster declarative and procedural knowledge; AI chatbots show promising effects on self-regulation and task performance. However, the evidence base is methodologically constrained: only five randomized experimental studies were identified. Across applications, AI is predominantly implemented through behaviorist or cognitively oriented instructional designs that emphasize drill-and-practice and adaptive feedback. In contrast, approaches fostering learner agency, critical reflection, and autonomous decision-making remain underrepresented. Conclusion Current research largely reflects a generalized „success narrative” surrounding AI in VET. Future studies should investigate failure cases, contextual moderators, and boundary conditions more systematically to develop a more differentiated understanding of the effectiveness of AI interventions. To realize the transformative potential of AI in VET, research and practice must move beyond replicating human instruction—avoiding the Turing Trap—and instead design learning environments that augment human judgment, strengthen learner agency, and support teachers in empathetic and holistic guidance.
- PaperDOAJ — Language assessment1 Jun 2026
Robust English Knowledge Tracing via Profile-Driven Forgetting and Masked Consistency
Xibo Chen, Ziqi Zhang, Haize Hu, Jie Jin et al.
A new framework for knowledge tracing in English learning uses a learning-profile-driven adaptive forgetting mechanism to model individual memory retention patterns and a masked consistency regularization method to reduce noise overfitting. Experiments show it outperforms state-of-the-art models in accuracy and robustness, supporting personalized language tutoring.
Original abstract
Knowledge Tracing (KT) plays a pivotal role in Intelligent Tutoring Systems (ITS) by dynamically assessing learners’ evolving knowledge states. However, tracking the acquisition of English presents unique challenges. Existing KT models typically employ homogeneous, predefined forgetting mechanisms that fail to capture the highly individualized nature of linguistic memory retention. Furthermore, language assessment data is notoriously noisy, which leads models to overfit superficial performance rather than capturing true underlying linguistic competence. To address these issues, we propose a novel framework to robustly trace English language competence. First, we introduce a Learning-Profile-Driven Adaptive Forgetting mechanism. Unlike methods with shared forgetting rates, our approach constructs a dynamic and strictly causal profile from historical interactions to generate personalized cognitive parameters (e.g., individualized forgetting rates). These parameters synchronously modulate the decay of multi-level knowledge states, enabling the model to accurately capture the heterogeneous memory retention patterns of different learners. Second, we design a Masked Consistency Regularization training paradigm. By applying stochastic masking to historical responses and enforcing predictive consistency, we prevent the model from exploiting localized noise and “shortcut” learning, compelling it to mine robust and invariant language representations. Extensive experiments on real-world educational datasets demonstrate that our proposed framework significantly outperforms state-of-the-art baselines in both prediction accuracy and noise resistance, offering a robust and interpretable solution for personalized language learning.