AI-mediated feedback in gamified programming education: effects on vocational students’ achievement and motivation

Abstract

This study examined the effects of an AI-enhanced gamified learning platform on vocational students’ programming achievement and interaction-related motivation. The platform was designed based on the GAFCC instructional model and incorporated real-time AI-mediated feedback to support learners during Java programming tasks. A quasi-experimental design was conducted with 90 first-year vocational students assigned to either an AI-enhanced gamified condition or a gamified condition without AI support. Programming achievement and interaction-related motivation were assessed following the intervention. Statistical analyses indicated that students in the AI-enhanced condition achieved significantly higher programming performance and reported stronger interaction-related motivation, particularly in confidence and satisfaction. The findings suggest that AI-mediated feedback is associated with improvements in learner–system interaction quality, which in turn supports better task performance and motivational responses.

1 Introduction

Growing demand for computational skills has led countries to prioritize programming education in vocational training. In China, policies such as the Digital China strategy and the Double High Plan have advanced the integration of programming and digital literacy into vocational curricula (Chen, 2020; Ministry of Education of China, 2025). As programming becomes central to vocational digital literacy, scalable and effective teaching strategies are increasingly needed.

These vocational education reforms aim to align training with industry needs across intelligent manufacturing, software services, and AI-driven operations (UNESCO, 2024). Game-based learning and gamification approaches have been widely recognized for enhancing learner motivation and engagement through structured game elements such as challenges, feedback, and rewards (Dumas Reyssier et al., 2024; Zourmpakis et al., 2023). By gamifying instructional tasks, GBL makes learning more active and supports cognitive development through scaffolded, immersive experiences (Gomez et al., 2023).

Adaptive gamification offers personalized feedback and task adjustments, but many systems still depend on static rules and fixed learner profiles. This limits their ability to respond to real-time changes in cognition, emotion, or behavior and reduces instructional flexibility (Bennani et al., 2022). Artificial Intelligence (AI) offers new opportunities to overcome these constraints. AI technologies such as machine learning, user modeling, and learning analytics enable real-time adaptation, personalized feedback, and intelligent task sequencing (Liu, 2025). Emerging research has begun to explore integrating AI-driven feedback into gamified learning environments to support real-time adaptation and personalized instruction (Kassenkhan et al., 2026; Montella et al., 2024).

However, empirical evidence in vocational programming contexts is still scarce. Despite these developments, existing studies often treat AI tutoring and gamification separately, with limited investigation of their combined effects on engagement and achievement. Vocational students often have limited prior knowledge, low metacognitive awareness, and fluctuating motivation, making programming tasks especially difficult (Mangaroska et al., 2022). These challenges require instructional strategies that are both responsive and individualized. AI-enhanced gamification could provide such support but remains underexplored in vocational contexts (Daghestani et al., 2020).

This study aims to examine how AI-mediated feedback shapes learner interaction in a gamified programming environment and to evaluate its effects on task performance and interaction-related motivation. Two complementary frameworks guide the development. Technologically, it applies the AIED framework to deliver adaptive instruction and real-time feedback (Hwang et al., 2020; Tapalova and Zhiyenbayeva, 2022). Pedagogically, the platform adopts the GAFCC model, comprising Goal, Attention, Feedback, Challenge, and Consolidation, to structure cognitively aligned game-based learning. Although the GAFCC model has yielded positive results in other domains, its use in vocational programming education remains underexplored.

From a learning-theory perspective, AI-mediated feedback can be viewed as adaptive scaffolding that supports learners’ progress in complex programming tasks. Consistent with constructivist learning theory, knowledge is actively constructed through interaction and problem-solving, while immediate AI feedback may reduce cognitive load and support incremental understanding, particularly for novice learners who often struggle with debugging and algorithmic reasoning. Building on this perspective, the present study examines how AI-mediated feedback within a gamified programming environment influences vocational students’ learning outcomes and interaction-related motivation. By linking adaptive AI feedback with established instructional design principles, this study extends existing research by integrating AI-mediated feedback with a structured gamified instructional model (GAFCC) and by examining both programming achievement and multidimensional interaction-related motivation within a unified framework.

2 Literature review

This review synthesizes recent studies on programming education in vocational contexts, focusing on learner challenges, gamification strategies, and the role of artificial intelligence in personalized learning. Programming is widely perceived as difficult by vocational students, who often lack prior experience in abstract reasoning and problem-solving. Common challenges include understanding syntax, applying algorithms, and sustaining learning motivation, which are further intensified by cognitive overload, limited interaction, and low learner confidence (Mangaroska et al., 2022; Zhan et al., 2022).

To address these challenges, gamification has been increasingly adopted as a learner-centered instructional approach. By integrating goals, feedback, rewards, and progress indicators, gamified learning environments have been shown to enhance learner motivation, engagement, and persistence in programming education (Nguyen-Viet et al., 2025). However, traditional gamification typically applies uniform game elements to all learners without accounting for individual differences in prior knowledge or learning progress. Such one-size-fits-all designs often fail to support diverse learner needs, leading to disengagement among advanced learners and frustration among beginners (Ekici, 2021).

Recent research, therefore, proposes adaptive gamification as an improvement over traditional approaches. Adaptive gamification adjusts task difficulty, feedback, and rewards based on learner performance and progress, and has demonstrated positive effects on engagement and learning outcomes (Dumas Reyssier et al., 2024; Zourmpakis et al., 2023). Nevertheless, many adaptive systems rely on predefined rules and static learner profiles, which limit real-time responsiveness and restrict their ability to capture dynamic changes in learners’ cognitive and motivational states (Şevk and Kocadere, 2024).

Artificial Intelligence in Education (AIED) has thus emerged as a promising direction for advancing adaptive gamified learning. Through learning analytics, user modeling, and intelligent agents, AI enables continuous monitoring of learner behavior and supports real-time adaptation of content, feedback, and task difficulty. Compared with rule-based approaches, AI-driven gamification offers greater flexibility and scalability, making it better suited to complex vocational learning environments. AI-enhanced adaptive gamification has been increasingly examined as a means to integrate real-time feedback, learner modeling, and generative AI for personalized learning pathways and dynamic instructional adjustment (Gómez Niño et al., 2025).

To guide the integration of gamification and AI, this study adopts the GAFCC model as an instructional framework. GAFCC provides a structured sequence for supporting learner engagement and instructional decision-making and has been applied in various educational contexts (Baah et al., 2023; Yang et al., 2023). However, existing applications of GAFCC often rely on static instructional plans with limited adaptive support. Although prior studies have reported positive effects of adaptive gamification and AI-based instruction on motivation and engagement, few systems integrate a structured pedagogical model with real-time AI-driven adaptation. This gap is particularly evident in vocational education, where learners frequently demonstrate low self-regulation, unstable motivation, and difficulty with abstract programming concepts (Kasneci et al., 2023). Addressing this gap, the present study develops and evaluates an AI-enhanced gamified learning platform grounded in the GAFCC framework for vocational programming education.

Figure 1

In line with the conceptual framework, Figure 2 presents the AI-mediated interaction mechanism that operationalizes the proposed model. The figure illustrates how learner input is processed to generate adaptive feedback. This mechanism provides the analytical basis for understanding how AI-mediated interaction supports both programming achievement and interaction-related motivation, as examined in the study. The mechanism involves three core stages: intent recognition, generative processing, and interactive feedback delivery.

Figure 2

Learner input is first interpreted through intent recognition and learner modeling, enabling adaptive instructional decisions (Hwang et al., 2020; Tapalova and Zhiyenbayeva, 2022). It is then processed to generate context-sensitive responses and scaffolded guidance (Montella et al., 2024), followed by real-time feedback to support task completion and sustained engagement. This interaction process offers a theoretical explanation for the relationship between AI-mediated feedback and both programming achievement and interaction-related motivation (Ng et al., 2024).

The interaction is completed through feedback, including explanations, task guidance, and motivational support. These functions correspond to the feedback and challenge components of the GAFCC model and directly support key motivational dimensions. Specifically, adaptive feedback enhances attention by maintaining learner focus, improves relevance by linking tasks to learner needs, and supports confidence through scaffolded guidance. In addition, continuous feedback and task progression contribute to satisfaction by reinforcing a sense of achievement. From a cognitive perspective, real-time feedback reduces task difficulty and supports problem-solving, thereby improving programming performance. This process forms a continuous cycle of interaction and adaptation aligned with AIED principles, providing a conceptual basis for understanding the relationship between AI-mediated interaction and both programming achievement and interaction-related motivation (Ng et al., 2024).

3 Materials and methods

This study aimed to examine the effects of AI-assisted gamified learning on vocational students’ programming achievement and interaction-related motivation. The primary research question was whether embedding real-time AI support within a game-based learning environment would improve learning outcomes compared with the same gamified platform without AI assistance. It was hypothesized that students in the AI-supported condition would demonstrate higher programming achievement and stronger interaction-related motivation.

A quasi-experimental design was employed. Participants were assigned to groups based on existing class arrangements. The experimental group received gamified programming instruction with real-time AI-supported feedback, while the comparison group used the same instructional content and game structure without AI support. Both groups were exposed to identical learning objectives, tasks, and instructional materials, ensuring that AI support was the only systematic difference between conditions. All instructional materials were developed using an interactive authoring platform and aligned with national curriculum standards for vocational programming education (Chiu et al., 2024; Labadze et al., 2023).

3.1 Instructional design framework and learning materials

The instructional design and delivery in this study were guided by the GAFCC model, which includes five pedagogical stages: goal setting, attention focusing, feedback, challenge, and consolidation. The study was conducted over a four-week period during regular classroom sessions. In Week 1, a pre-test was administered, followed by a two-week intervention phase (Week 2–3). In Week 4, a post-test and questionnaire were conducted. All learning materials were developed using Articulate Storyline 360 and aligned with the 2025 Higher Vocational Software Technology Standards issued by the Ministry of Education of China.

The gamified learning environment was structured using a board-game-style interface to guide learners through the instructional sequence (Figure 3). The interface organizes learning tasks into progressive stages and provides clear goals at the beginning of each unit. This design reflects the task sequencing and goal-orientation components of the GAFCC model. Visual progression and structured task flow support sustained attention during programming activities. Embedded multimedia elements, such as instructional videos and interactive quizzes, further maintain engagement and provide contextual learning support. This structure supports attention and relevance by providing clear task organization and continuous goal progression within the learning sequence.

Figure 3

Formative feedback was delivered through real-time AI-assisted responses during learning tasks. When learners encountered errors or uncertainties, they could request immediate support through the AI interface. The system provided context-specific error explanations and corrective guidance. This feedback function acts as adaptive scaffolding, helping learners identify mistakes and refine their solutions during task execution (Figure 4). It reduces uncertainty and supports learner confidence, thereby promoting continuous engagement with programming tasks.

Figure 4

Programming challenges were designed as structured coding activities that increased in difficulty. Learners applied Java programming concepts in an interactive coding environment and received AI-assisted guidance when needed. Step-by-step hints and error diagnostics were provided to support problem-solving and iterative improvement. Key programming concepts were revisited across multiple tasks to strengthen knowledge consolidation (Figure 5). This structure supports satisfaction through successful task completion and contributes to improved programming performance.

Figure 5

Throughout the instructional process, learner interactions were automatically recorded. These data supported adaptive feedback during instruction and enabled subsequent analysis of learning performance and interaction patterns. This data collection also provided an empirical basis for evaluating the effectiveness of the AI-mediated gamified learning environment.

3.2 AI-supported mechanism

The generative AI component was implemented using a large language model accessed via an API interface. Prompt templates were predefined to structure learner inputs, incorporating task context, error type, and expected learning objectives. This design ensured consistent feedback generation across learners and tasks. Feedback generation followed a hybrid approach combining rule-guided adaptation with generative AI-based responses, where predefined rules mapped learner performance conditions to corresponding feedback strategies. Repeated errors prompted more explicit hints, while correct or near-correct responses elicited concise feedback (Kasneci et al., 2023).

As shown in Figure 6, the AI-supported learning mechanism comprises three layers: the data layer, the generative AI layer, and the output layer. The data layer collects learner activity, conversation history, and question bank resources to support user modeling. The generative AI layer processes structured prompts using a template-based prompt builder and a large-language-model API. Learner input is first formatted into predefined prompt structures. It is then analyzed to identify error patterns. Based on this analysis, feedback strategies are selected and adapted according to predefined rules and learner performance conditions. The output layer provides real-time instructional support, including dynamic questions, personalized learning pathways, and motivational feedback. This layered design ensures systematic processing of learner input and supports adaptive feedback and formative assessment for personalized learning (Ng et al., 2024).

Figure 6

3.3 Participants and sampling

Participants were first-year vocational students from a vocational college in Southeast China, enrolled in an introductory programming course. A total of 90 students participated in the study, with 45 in the experimental group and 45 in the comparison group. Participants were assigned to groups based on intact class structures, consistent with a quasi-experimental design, as random assignment was not feasible in the natural classroom setting. Both groups received instruction on the same content and learning objectives within the four-week study period. The instructional intervention was implemented over two weeks (Week 2–3) during regular classroom sessions, ensuring equivalent instructional exposure across groups. The instructional content was based on the textbook Java Programming: Fundamentals and Practice, published by People's Posts and Telecommunications Press, which aligns with the Higher Vocational Software Technology Standards issued by the Ministry of Education of China (2025), ensuring curricular consistency across groups.

3.4 Research instruments

The design focuses on evaluating two key learning outcomes: programming achievement and interaction-related motivation. Programming achievement was measured using the Computer Programming Achievement Test (CPAT), developed in accordance with national curriculum standards. Interaction-related motivation was assessed using the Instructional Materials Motivation Survey (IMMS), grounded in the ARCS motivational model (Keller, 2010). Both instruments have been widely used in programming and instructional design research. Pre-test and post-test scores were analyzed using analysis of covariance (ANCOVA) to control for baseline differences. This provides evidence supporting the effectiveness of the intervention and aligns with established guidelines for interpreting ANCOVA in educational research (Field, 2018).

In this study, interaction-related motivation refers to learners’ motivational responses elicited through interaction with the AI-mediated gamified system. It reflects how system feedback, task adaptation, and interface features influence learners’ attention, perceived relevance, confidence, and satisfaction during programming tasks. These dimensions were operationalized using the ARCS-based Instructional Materials Motivation Survey (IMMS). In the main study sample, the IMMS demonstrated good internal consistency, with Cronbach's alpha coefficients ranging from.85 to.90 across the four dimensions. Previous studies have also reported strong reliability of the IMMS in programming and digital learning contexts (Robertson and Evans, 2020), supporting its use for measuring interaction-related motivation based on the ARCS model (Keller, 2010).

The CPAT was developed in accordance with national curriculum standards and reviewed by three programming education experts to ensure content validity, with a focus on item relevance, curriculum alignment, and coverage of key programming concepts. In the main study, the CPAT achieved a Kuder–Richardson Formula 20 (KR-20) reliability coefficient of 0.78, indicating acceptable internal consistency. These results support the suitability of the CPAT for evaluating programming achievement in the present study.

3.5 Experimental procedures

The experiment lasted 4 weeks, with a 2-week intervention phase (Weeks 2–3). In the first week, all participants completed a 60 min pre-test using the Computer Programming Achievement Test (CPAT) to assess their basic knowledge. In the second and third weeks, both groups of participants engaged in 80 min of gamified learning activities each week. The experimental group (AI-GAME) used a gamified learning system enhanced with artificial intelligence assistance. In contrast, the control group (GAME) used the same gamified interface but received only fixed, non-adaptive feedback. Both groups had access to the same content, including video lessons, quizzes, and debugging tasks. In the fourth week, all students completed an 80-minute post-test (CPAT) and an IMMS questionnaire to assess their learning motivation. Figure 7 provides a visual overview of the experimental process, including the grouping, intervention, and assessment phases.

Figure 7

To maintain internal validity, all participants followed the same instructional sequence, content, game elements, and learning duration. The only manipulation was the use of AI-based instructional support, which allowed for a comparative analysis of its impact on learning performance and motivation (Ng et al., 2024). The GAFCC model provided a common instructional framework for both groups of students, which the AI-supported system extended with real-time feedback and adaptive scaffolding. Based on this controlled design, the data analysis results are presented in the following sections.

4 Results

This section presents the key findings from the study's quantitative analysis. It compares the instructional effects of the two gamified learning platforms, AI-GAME and GAME, on vocational students’ programming achievement and interaction-related motivation. Results are organized by outcome variables: achievement (measured by CPAT) and motivation (measured by IMMS across the four ARCS dimensions: Attention, Relevance, Confidence, and Satisfaction). The data are analyzed using ANCOVA and one-way ANOVA where appropriate. Descriptive statistics, statistical significance, and observed trends are reported to support interpretation.

4.1 Learning achievement

The Computer Programming Achievement Test (CPAT), developed by the researcher, was used to assess students’ programming achievement. It consisted of 30 items: 20 multiple-choice questions (each worth 2 marks) and 10 fill-in-the-blank questions (each worth 6 marks), all aligned with the core content of the vocational Java programming syllabus. Content validity was established through expert review, and item analysis was conducted to examine difficulty and discrimination indices. The Kuder–Richardson Formula 20 (KR-20) was used to assess internal consistency, yielding a coefficient of.84 based on pilot data (N = 35), indicating good reliability for educational testing (Ntumi et al., 2023).

The ANCOVA revealed a significant main effect of group on post-test performance, F (1, 87) = 40.31, p < .001, partial η² = 0.32. After adjusting for pre-test scores, the AI-GAME group achieved a higher mean score (M = 85.96) compared to the GAME group (M = 73.62), indicating a substantial instructional effect. The adjusted means were similar to the raw means, suggesting that baseline knowledge had minimal impact on the outcomes. This provides evidence supporting the effectiveness of the intervention and aligns with prior recommendations for interpreting ANCOVA under conditions of minimal covariate influence (Miyazaki et al., 2022; Field, 2018). Results are summarised in Table 1.

A one-way ANCOVA was conducted to compare post-test scores between groups, with pre-test scores as the covariate. Preliminary assumption checks supported the use of ANCOVA. The Shapiro–Wilk test confirmed normality for the GAME group (p = .269). Although the AI-GAME group showed slight non-normality (p = .021), the sample sizes were equal, and ANCOVA is generally considered robust to moderate violations of normality. Levene's test indicated homogeneity of variance (p = .773). The interaction between group and covariate was not statistically significant (p = .139), confirming the assumption of homogeneity of regression slopes. These results justified the application of ANCOVA.

4.2 Interaction-related motivation

A customized version of the Instructional Materials Motivation Survey (IMMS; Keller, 2010) was used to measure interaction-related motivation in the AI-GAME and GAME groups. The instrument followed the ARCS model and included four dimensions: attention, relevance, confidence, and satisfaction, with 36 items in total. Responses were rated on a five-point Likert scale (1 = not true, 5 = very true). A pilot study (N = 35) indicated good internal consistency, with Cronbach's alpha coefficients ranging from.83 to.88 across the four dimensions. Item-level ANOVA results for the IMMS dimensions are reported in the table. Overall, the AI-GAME group reported higher motivation scores than the GAME group across all four ARCS dimensions, with several items showing statistically significant differences.

Item-level analyses further supported these findings, with several items within each dimension showing statistically significant differences in favor of the AI-GAME group (Tables 2–5). Specifically, significant differences were observed in attention items (Q1, Q2, Q10), relevance items (Q14, Q18, Q19), and a confidence-related item (Q22) (p < .05). These results indicate that AI-mediated support had a stronger effect on feedback-driven, interaction-based, and application-oriented aspects of motivation, which are closely related to real-time interaction and adaptive guidance. Following the item-level findings, construct-level comparisons were conducted using ANCOVA to examine differences across the four ARCS dimensions, while controlling for pre-test programming achievement (Table 6).

The results showed a significant effect of instructional group on all four dimensions. For attention, the AI-GAME group scored higher than the GAME group, F(1, 87) = 52.04, p < .001, with a large effect size (partial η² = 0.37). For relevance, the AI-GAME group also outperformed the GAME group, F(1, 87) = 58.65, p < .001, with a large effect size (partial η² = .40). For confidence, a significant difference was found in favour of the AI-GAME group, F(1, 87) = 8.95, p = .004, with a moderate effect size (partial η² = 0.09). For satisfaction, the AI-GAME group again showed higher scores, F(1, 87) = 23.56, p < .001, with a large effect size (partial η² = 0.21). Across all dimensions, the adjusted mean scores consistently favoured the AI-GAME condition.

The pattern of effect sizes indicates that the greatest improvements occurred in attention and relevance, while confidence showed a smaller but meaningful effect. The covariate (pre-test programming achievement) was not statistically significant across all models, suggesting that baseline differences did not substantially influence the results and that the observed effects were attributable to the instructional intervention. Overall, the construct-level analysis provides a more robust and comprehensive interpretation, reducing the risk of inflated Type I error associated with multiple item-level comparisons.

5 Discussion

The findings indicate that the AI-mediated condition was associated with higher programming achievement than the gamified-only condition. For interaction-related motivation, the results partially support H02a–H02d, with significant differences across all ARCS dimensions, particularly in attention and relevance. Overall, the AI-GAME group demonstrated higher performance and motivation than the GAME group, suggesting that integrating AI support within a gamified learning environment is associated with improved learning outcomes. These findings are consistent with prior research indicating that AI-supported gamification is associated with enhanced learner engagement and performance in adaptive learning contexts (Dumas Reyssier et al., 2024; Ng et al., 2024). Given that both groups shared identical gamified structures, the observed differences can be attributed primarily to the presence of AI-mediated feedback rather than gamification alone.

From a theoretical perspective, these findings may be interpreted in terms of adaptive scaffolding in learning. Consistent with constructivist learning theory, knowledge is actively constructed through interaction and problem-solving, and AI-mediated feedback can be viewed as adaptive scaffolding that supports iterative learning and cognitive development (Kassenkhan et al., 2026). In this context, real-time feedback may assist learners by providing immediate guidance, error correction, and task-specific hints, which help refine understanding during coding activities. This scaffolding process also aligns with cognitive load theory, as such support may reduce unnecessary cognitive burden by directing attention to relevant problem-solving processes rather than syntax or debugging difficulties.

Consistent with the ANCOVA results (Table 6), the strongest effects were observed in attention and relevance. These findings suggest that AI-supported feedback may contribute to sustaining learner attention and perceived task value. Immediate and context-sensitive feedback may help maintain focus during complex programming tasks, while personalized guidance may enhance the perceived usefulness of learning activities. This interpretation is consistent with prior research indicating that adaptive, AI-enhanced gamified environments are associated with improved learner engagement and motivation (Dumas Reyssier et al., 2024; Ng et al., 2024).

In contrast, improvements in confidence and satisfaction were smaller. This may be explained by learners’ limited familiarity with AI-supported learning environments and the trial-and-error nature of programming tasks. While AI feedback supports task completion, building confidence and satisfaction may require longer exposure and repeated successful experiences. This suggests that deeper motivational dimensions develop more gradually compared with immediate engagement-related responses.

Traditional programming instruction in vocational education often relies on static instructional materials and delayed feedback, which may limit learners’ opportunities to refine problem-solving strategies during practice. Gamified learning environments have been introduced to address these limitations by increasing learner engagement and providing structured practice opportunities. Through game elements such as goals, challenges, and progressive tasks, gamified instruction can encourage active participation and sustained practice, consistent with meta-analytic evidence showing that gamification improves programming learning outcomes and engagement compared with traditional instruction (Zhan et al., 2022). In the present study, the addition of AI support was associated with further improvements in motivation and achievement outcomes, as reflected in the higher scores observed in the AI-GAME condition (Ng et al., 2024).

However, the present study focused on outcome variables, including programming achievement and interaction-related motivation, and did not directly measure underlying cognitive or interaction processes. Therefore, interpretations of mechanisms such as adaptive scaffolding, cognitive load reduction, or interaction quality should be considered inferential rather than empirically verified. Future research should incorporate process-level data, such as interaction logs, cognitive load measures, or learning analytics, to provide a more comprehensive understanding of how AI-supported gamification influences learning.

In addition to the theoretical contributions, this study offers practical implications for teaching by highlighting the value of integrating AI-mediated feedback into gamified programming instruction. In practice, teachers may use AI tools to provide immediate feedback during coding tasks and support real-time error correction. It is also beneficial to align AI support with structured game elements, such as goal setting and progressive challenges, to maintain engagement. In this way, AI functions as a supplementary instructional tool that enhances personalized learning rather than replacing teacher guidance. Overall, these findings highlight the value of integrating AI-supported feedback to enhance programming achievement and motivation.

6 Limitations and future research

Despite the significant findings, several limitations should be considered when interpreting this study’s results.

First, the duration of the intervention was relatively short (two weeks), which may have introduced a novelty effect, characterized by increased initial engagement due to the newness of the learning environment. Prior studies suggest that short-term gamified interventions may overestimate motivational effects due to such initial engagement (Zhan et al., 2022). Therefore, the observed improvements in motivation and achievement may partly reflect learners’ initial engagement with the AI-supported environment rather than sustained learning effects. Future research should adopt longer intervention periods to examine the persistence of these effects over time.

Second, the sample comprised 90 vocational students from a single institution, which may limit the generalizability of the findings to broader populations. Prior research suggests that the effects of AI-enhanced gamification may vary across educational contexts and learner characteristics (Marengo et al., 2025). Future research should include larger and more diverse samples across multiple institutions to strengthen the external validity of the results.

Third, this study did not explicitly control for students’ digital literacy or prior technology experience. Differences in learners’ familiarity with digital tools or AI systems may have influenced how they interacted with the platform and processed AI-generated feedback. Students with lower digital literacy may have experienced difficulty interpreting feedback, which could reduce confidence and satisfaction. Future studies should include digital literacy as a control or moderating variable.

Finally, the study was conducted within the context of vocational education in China. Educational practices, curriculum structures, and learner characteristics in this context may differ from those in other regions or educational systems. Therefore, caution is needed when generalizing the findings to other cultural or institutional settings. Future research should examine the applicability of AI-enhanced gamified learning in different educational contexts.

In addition, this study focused on outcome variables and did not directly measure underlying cognitive or interaction processes. Future research should incorporate process-level data, such as interaction logs, cognitive load measures, or learning analytics, to provide deeper insights into the mechanisms of AI-supported learning, as recent reviews emphasize the importance of such data for understanding how AI-driven gamification is associated with learning processes (Hsiao et al., 2023). Given these limitations, the GAFCC model may offer a flexible instructional framework beyond the present setting. Its core components, including goal setting, attention support, feedback, challenge, and consolidation, are grounded in general instructional design and motivational principles and may be applicable across different learning domains. However, its effectiveness in diverse subjects and cultural contexts requires further empirical validation.

References

• 1

  BaahC.GovenderI.Rontala SubramaniamP. (2023). Exploring the role of gamification in motivating students to learn. Cogent Educ.10 (1), 2210045. 10.1080/2331186X.2023.2210045

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  BennaniS.MaalelA.Ben GhezalaH. (2022). Adaptive gamification in E-learning: a literature review and future challenges. Comput. Appl. Eng. Educ.30 (2), 628–642. 10.1002/cae.22477

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  ChenC.-H. (2020). Impacts of augmented reality and a digital game on students’ science learning with reflection prompts in multimedia learning. Educ. Technol. Res. Dev.68 (6), 3057–3076. 10.1007/s11423-020-09834-w

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  ChiuT. K.MoorhouseB. L.ChaiC. S.IsmailovM. (2024). Teacher support and student motivation to learn with artificial intelligence (AI) based chatbot. Interact. Learn. Environ.32 (7), 3240–3256. 10.1080/10494820.2023.2172044

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  DaghestaniL. F.IbrahimL. F.Al-TowirgiR. S.SalmanH. A. (2020). Adapting gamified learning systems using educational data mining techniques. Comput. Appl. Eng. Educ.28 (3), 568–589. 10.1002/cae.22227

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  Dumas ReyssierS.SernaA.HallifaxS.MartyJ.-C.SimonianS.LavouéE. (2024). How does adaptive gamification impact different types of student motivation over time?Interact. Learn. Environ.32 (10), 6043–6062. 10.1080/10494820.2023.2248220

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  EkiciM. (2021). A systematic review of the use of gamification in flipped learning. Educ. Inf. Technol.26 (3), 3327–3346. 10.1007/s10639-020-10394-y

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  FieldA. (2018). Discovering Statistics Using IBM SPSS Statistics. 5th Ed. London: SAGE Publications.

  • Google Scholar
  • View reference in article

• 9

  GomezM. J.Ruipérez-ValienteJ. A.ClementeF. J. G. (2023). A systematic literature review of game-based assessment studies: trends and challenges. IEEE Trans. Learn. Technol.16 (4), 500–515. 10.1109/TLT.2022.3226661

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  Gómez NiñoJ. R.Árias DelgadoL. P.ChiappeA.Ortega GonzálezE. (2025). Gamifying learning with AI: a pathway to 21st-century skills. J. Res. Child. Educ.39 (4), 735–750. 10.1080/02568543.2024.2421974

  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  HsiaoH. S.ChangC. S.LinC. Y.WangY. Z. (2023). Effects of AI-based gamification learning environments on students' programming learning performance, motivation, and engagement. Comput. Educ. Artif. Intell.4, 100127.

  • Google Scholar

• 12

  HwangG.-J.XieH.WahB. W.GaševićD. (2020). Vision, challenges, roles and research issues of artificial intelligence in education. Comput. Educ.: Artif. Intell.1, 100001. 10.1016/j.caeai.2020.100001

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  IshaqK.AlviA.HaqM. I.RosdiF.ChoudhryA. N.AnjumA.et al (2024). Level up your coding: a systematic review of personalized, cognitive, and gamified learning in programming education. PeerJ Comput. Sci.10, e2310. 10.7717/peerj-cs.2310

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  KasneciE.SesslerK.KüchemannS.BannertM.DementievaD.FischerF.et al (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learn. Individ. Differ.103, 102274. 10.1016/j.lindif.2023.102274

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  KassenkhanA.SerbinV.BeisembekovaR.AbshukirovaA.MendekinaB. (2026). AI-Supported Gamification in E-learning: a systematic review of adaptive architectures and cognitive outcomes. Information17 (3), 282. 10.3390/info17030282

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  KellerJ. M. (2010). Motivational design for learning and performance: the ARCS model approach. US: Springer. 10.1007/978-1-4419-1250-3

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  LabadzeL.GrigoliaM.MachaidzeL. (2023). Role of AI chatbots in education: systematic literature review. Int. J. Educ. Technol. High. Educ.20 (1), 56. 10.1186/s41239-023-00426-1

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  LiuL. (2025). Impact of AI gamification on EFL learning outcomes and nonlinear dynamic motivation: comparing adaptive learning paths, conversational agents, and storytelling. Educ. Inf. Technol.30 (8), 11299–11338. 10.1007/s10639-024-13296-5

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  MangaroskaK.SharmaK.GaševićD.GiannakosM. (2022). Exploring students’ cognitive and affective states during problem solving through multimodal data: lessons learned from a programming activity. J. Comput. Assist. Learn.38 (1), 40–59. 10.1111/jcal.12590

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  MarengoA.PaganoA.LundB.SantamatoV. (2025). Research AI: integrating AI and gamification in higher education for e-learning optimization and soft skills assessment through a cross-study synthesis. Front. Comput. Sci.7, 1587040. 10.3389/fcomp.2025.1587040

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  Ministry of Education of the People's Republic of China. (2025). Electronics and Information Category. Beijing: Ministry of Education of the People's Republic of China.Avialiable online at: http://www.moe.gov.cn/s78/A07/zcs_ztzl/2017_zt06/17zt06_bznr/bznr_zyjyzyjxbz/gdzyjy_zk/zk_dzyxxdl/(Accessed May 6 2026).

  • Google Scholar
  • View reference in article

• 22

  MiyazakiY.KamataA.UekawaK.SunY. (2022). Bias for treatment effect by measurement error in pretest in ANCOVA analysis. Educ. Psychol. Meas.82 (6), 1130–1152.

  • Pubmed Abstract
  • Google Scholar

• 23

  MontellaR.De VitaC. G.MelloneG.CiricilloT.CaramielloD.Di LuccioD.et al (2024). Leveraging large language models to support authoring gamified programming exercises. Appl. Sci.14 (18), 1–15. 10.3390/app14188344

  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  NgD. T. K.XinyuC.LeungJ. K. L.ChuS. K. W. (2024). Fostering students’ AI literacy development through educational games: AI knowledge, affective and cognitive engagement. J. Comput. Assist. Learn.40 (5), 2049–2064. 10.1111/jcal.13009

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  Nguyen-VietB.Nguyen-DuyC.Nguyen-VietB. (2025). How does gamification affect learning effectiveness? The mediating roles of engagement, satisfaction, and intrinsic motivation. Interact. Learn. Environ.33 (3), 2635–2653. 10.1080/10494820.2024.2414356

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  NtumiS.AgbenyoS.BulalaT. (2023). Estimating the psychometric properties (‘item difficulty, discrimination and reliability indices’) of test items using Kuder-Richardson approach (KR-20). Shanlax Int. J. Educ.11 (3), 18–28. 10.34293/education.v11i3.6081

  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  ReyssierS.HallifaxS.SernaA.MartyJ.-C.SimonianS.LavouéE. (2022). The impact of game elements on learner motivation: influence of initial motivation and player profile. IEEE Trans. Learn. Technol.15 (1), 42–54. 10.1109/TLT.2022.3153239

  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  RobertsonO.EvansM. S. (2020). Just how reliable is your internal reliability? An overview of cronbach’s alpha (α). PsyPag Q.1 (115), 23–27. 10.53841/bpspag.2020.1.115.23

  • CrossRef
  • Google Scholar
  • View reference in article

• 29

  ŞevkK. E.KocadereS. A. (2024). A review on personalization of gamified learning environments. J. Theor. Educ. Sci.17 (4), 873–896. 10.30831/akukeg.1478810

  • CrossRef
  • Google Scholar
  • View reference in article

• 30

  TapalovaO.ZhiyenbayevaN. (2022). Artificial intelligence in education: AIEd for personalised learning pathways. Electron. J. E-Learn.20 (5), 639–653. 10.34190/ejel.20.5.2597

  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  UNESCO (2024). New UNESCO Report on TVET Recommendations to Improve Global Skills Development. Paris: UNESCO. Available online athttps://www.unesco.org/en/articles/new-unesco-report-tvet-recommendations-improve-global-skills-development(Accessed May 6, 2026).

  • Google Scholar
  • View reference in article

• 32

  YangQ.-F.LianL.-W.ZhaoJ.-H. (2023). Developing a gamified artificial intelligence educational robot to promote learning effectiveness and behavior in laboratory safety courses for undergraduate students. Int. J. Educ. Technol. High. Educ.20 (1), 18. 10.1186/s41239-023-00391-9

  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  ZhanZ.HeL.TongY.LiangX.GuoS.LanX. (2022). The effectiveness of gamification in programming education: evidence from a meta-analysis. Comput. Educ.: Artif. Intell.3, 100096. 10.1016/j.caeai.2022.100096

  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  ZourmpakisA.-I.KalogiannakisM.PapadakisS. (2023). Adaptive gamification in science education: an analysis of the impact of implementation and adapted game elements on students’ motivation. Computers12 (7), 143. 10.3390/computers12070143

  • CrossRef
  • Google Scholar
  • View reference in article