Rapid advances in Artificial Intelligence (AI), Digital Twin (DT) technology, and the Metaverse have created new opportunities for transforming STEM education and workforce preparation. However, existing learning systems remain fragmented and lack an integrated, human-centered framework capable of supporting high-fidelity simulations, adaptive learning, and real-time competence development aligned with Industry 5.0. This study addresses this gap by developing a unified AI-Based Digital Twins Metaverse (AI-DTM) framework aimed at strengthening STEM employability skills and fostering adaptive expertise in higher education. This research employed a conceptual–analytical design comprising three structured phases: (1) a systematic conceptual synthesis guided by PRISMA 2020 to identify technological, pedagogical, and human-centric mechanisms relevant to AI, DTs, and Metaverse-based learning; (2) construction of an integrated AI-DTM framework through theoretical mapping, concept modeling, and iterative refinement; and (3) expert validation involving specialists in AI systems, digital twins, immersive learning, and instructional design. Evidence from interdisciplinary literature across AI-enhanced learning, DT-enabled simulation, and XR-based experiential environments informed the development of the framework. The resulting AI-DTM framework integrates AI-driven learner modeling, real-time DT simulations, and immersive Metaverse environments to create a unified, adaptive ecosystem. Key outcomes include: (a) personalized learning pathways supported by intelligent analytics and automated feedback; (b) high-fidelity, risk-free simulations that replicate authentic STEM work processes; (c) immersive and collaborative virtual experiences that enhance engagement, problem-solving, and teamwork; and (d) continuous competence profiling enabling the development of adaptive expertise. The framework demonstrates strong alignment with Industry 5.0 principles, supporting human–AI collaboration, data-driven decision-making, and future workforce readiness. This study provides a novel, scalable model that advances the design of AI-enabled learning ecosystems by integrating AI, Digital Twins, and Metaverse technologies into a cohesive architecture. The AI-DTM framework offers theoretical and practical contributions for enhancing STEM employability skills, strengthening adaptive expertise, and guiding institutions.

Mathematics is a basic discipline which causes anxiety to primary school students, resulting in reduced motivation for learning and poor achievements. Traditional instruction heavily leans on abstract symbols, which can add to cognitive load and a feeling of emotional opposition. To address this issue, this study developed a learning model integrating augmented reality (AR) and haptic feedback to create an interactive and multisensory mathematics experience. A quasi-experimental design was conducted with 120 students divided into experimental and control groups. The experimental group used the AR–haptic mathematics game for six weeks, while the control group learned through textbooks and teacher explanations. Data was gathered by using the conceptual understanding tests, a mathematics anxiety scale, classroom observations and student interviews. The quantitative data were analyzed by t-test, ANCOVA and MANOVA analysis; qualitative data were analyzed positionally. The results indicated significant progress in experimental groups’ conceptual understanding and mathematics anxiety decrease when compared to the control group. The mean for the post-test experimental group was 82, and for the control group, it was 62 with a large effect size. The average level of maths anxiety decreased in the experimental group from 3.2 to 1.8 on a five-point scale. Qualitative feedback revealed that students found the game engaging, enjoyable, and less intimidating than traditional lessons. The integration of AR and haptic feedback effectively enhanced both cognitive and affective outcomes in mathematics learning. The multisensory environment enabled students to interact with abstract concepts in tangible ways, reducing anxiety and promoting positive attitudes toward mathematics. This provides a proof of concept for the effectiveness of immersive technologies as assistive tools in the enhancement of academic achievement and emotional health related to elementary level mathematics learning.

Vocational students often experience career path anxiety due to uncertainty about labor market demands, limited mentoring, and misalignment between curricula and industry needs. In Indonesia, this is amplified by uneven career guidance despite mandates for workforce readiness. Recent advances in artificial intelligence (AI) enable adaptive, data-driven, and psychologically informed support that links students’ skills with real-time labor markets. This study used a design science research approach to build and evaluate an AI-driven career guidance system with three components: (1) a supervised machine learning skills mapping engine, (2) an adaptive mentoring module using an AI chatbot and mentor matching, and (3) a real-time labor market intelligence module using natural language processing to analyze job postings and trends. A mixed-methods evaluation involved 180 vocational students from three schools in South Kalimantan assigned to intervention and control groups. Quantitative data were collected through pre–post career anxiety surveys and system performance metrics, while qualitative data were gathered through interviews and focus group discussions. Analysis included paired-sample t-tests, predictive model evaluation, and thematic analysis. Students using the AI system showed a significant 26.7% reduction in career path anxiety compared with minimal change in the control group (p < 0.001). The skills mapping model achieved 87% accuracy in predicting suitable career pathways with strong precision, recall, and F1-scores. Engagement was high: 65% repeatedly conducted skill-gap analyses, 79% joined adaptive mentoring, and 87% downloaded personalized career roadmaps. Qualitative findings revealed greater confidence, clearer direction, and better alignment between students’ competencies and labor market expectations. The AI-driven career guidance system effectively reduced career anxiety while strengthening readiness through personalized skills mapping, adaptive mentoring, and real-time labor market intelligence. The study shows that human-centered AI can enhance vocational guidance, bridge school–industry gaps, and support more confident, evidence-based career decisions among vocational students in Indonesia nationwide.

The integration of digital technology in vocational education has redefined instructional delivery and learner engagement. However, the extent to which technology enhances learning depends largely on the quality of teacher–student interaction. This study investigates the mediating role of Teacher–Student Interaction (TSI) in the relationship between the Technology-Enhanced Learning Environment (TEL) and Learning Engagement (LE). Drawing on the Community of Inquiry (CoI) and Technological Pedagogical Content Knowledge (TPACK) frameworks, this research tested a structural model to explain how interaction bridges technological and pedagogical processes in vocational learning contexts. A total of 362 valid responses were collected from vocational students across Indonesia using a stratified sampling approach. The data were analyzed through Structural Equation Modelling (SEM) using SmartPLS 4 with 5,000 bootstrap samples. The results demonstrated that TEL positively influenced TSI (β = 0.71, p < .001) and LE (β = 0.39, p < .01). Moreover, TSI significantly predicted LE (β = 0.53, p < .001). The indirect pathway from TEL to LE through TSI was also significant (β = 0.37, p < .001), indicating partial mediation with a Variance Accounted for (VAF) of 48.7%. The model achieved an excellent fit (χ 2 /df = 2.31, CFI = 0.953, TLI = 0.947, SRMR = 0.046, RMSEA = 0.048) and explained 55% of the variance in TSI and 68% in LE. These findings affirm that technology integration enhances learning engagement primarily when mediated by active teacher–student interaction. The study underscores that digital transformation in vocational education must emphasize pedagogical presence and communicative interaction to sustain engagement, particularly in remote and under-resourced (3T) regions.

Incorporating Indigenous knowledge into Technical and Vocational Education and Training (TVET) is recognized as one of the key strategies to support culturally responsive and inclusive education. In the sections that follow, we draw upon current international literature and studies to identify approaches with respect to embedding Indigenous knowledge into TVET curricula - strategies and challenges; outcomes. Eighty-five peer-reviewed articles between 2020 and 2025 were reviewed according to PRISMA 2020, concentrating on pedagogical models and methods, curriculum construction process, community involvement and student performance. The outcomes reveal that effective integration is grounded on culturally relevant teaching resources and active engagement with indigenous communities, along with traditional knowledge systems coupling to technical skills. Some of the challenges involve institutionalized barriers, a lack of teacher training and resources can restrict the ways in which we see Indigenous perspectives incorporated in meaningful ways. Notwithstanding these obstacles, there is evidence that students who experience Indigenous knowledge in TVET display higher levels of cultural awareness and identity formation, and engagement with learning; their communities are more empowered to participate in education. The review emphasizes the necessity for an overall framework and policies supporting sustainable integration, and advises that long-term effects on employability, social inclusion and community development be investigated in future studies. By highlighting the importance of Indigenous knowledge in TVET, this study adds to discussions on culturally-responsive pedagogical practice and offers practical advice for educators, policy makers and researchers aiming to improve TVET practices across the world.

Background Generation Alpha (born after 2010) requires pedagogical approaches aligned with their digital-native characteristics and Society 5.0 demands. Although digital integration advances in curriculum, many Indonesian vocational schools persist with teacher-centered methods that prioritize technical skills while neglecting essential soft skills teamwork, creativity, and adaptability. This study developed and validated the Alpha Generation Learning Style Scale (ALSS), the first instrument measuring Generation Alpha’s integrated learning preferences in digital vocational settings. Grounded in social constructivism, connectivism, heutagogy, and cybergogy, ALSS comprises four interconnected dimensions: Visual-Digital Learning (VDL), Collaborative Learning (CL), Self-Directed Learning (SDL), and Gamified Learning (GL). Methods Using Cantabrana et al.’s (2019) four-phase framework, the study executed expert validation (n=5), pre-testing (n=35), and empirical testing with 285 stratified vocational students using Classical Test Theory and Confirmatory Factor Analysis (CFA). Results ALSS demonstrated excellent content validity (S-CVI/Ave = 0.981; Aiken’s V > 0.85) and high internal consistency (Cronbach’s α = 0.96). CFA confirmed a strong second-order factor structure (CFI = 0.996; TLI = 0.996; RMSEA = 0.020; SRMR = 0.030), with all dimensions significantly loading onto ALSS (λ = 0.59–0.65, p < 0.001). Self-Directed Learning showed the strongest influence (0.645), followed by Collaborative Learning (0.640) and Visual-Digital Learning (0.627), explaining 35–42% of variance per dimension. Convergent validity (AVE = 0.821–0.864), discriminant validity, and composite reliability (CR > 0.96) were confirmed. Conclusions ALSS is a valid, reliable instrument capturing Generation Alpha’s technology-mediated learning identity. It provides an evidence-based diagnostic framework for personalizing Cooperative Project-Based Learning (Co-PjBL) and transforming Indonesian vocational education toward adaptive, student-centered methods. Integration with computational thinking skills including algorithmic thinking, decomposition, pattern recognition, and abstraction strengthens pedagogical effectiveness. This approach aligns TVET with Society 5.0 demands, enhancing preparation of digitally proficient, computationally literate, collaborative, and autonomous learners.