<![CDATA[Perkembangan kecerdasan artifisial (artificial intelligence/AI) memperluas cara pembelajaran Science, Technology, Engineering, and Mathematics (STEM) dirancang, dilaksanakan, dan dinilai. Kajian ini bertujuan mensintesis bentuk aplikasi AI, praktik pedagogis, hasil belajar, serta tantangan implementasinya dalam pembelajaran STEM. Kajian menggunakan pendekatan integrative literature review terstruktur terhadap 30 publikasi inti yang terbit pada 2019–2025. Literatur ditelusuri melalui metadata ilmiah, indeks pendidikan, dan laman penerbit menggunakan kombinasi kata kunci STEM education, artificial intelligence, generative AI, intelligent tutoring system, adaptive learning, learning analytics, educational robotics, dan learning outcomes. Artikel dipilih apabila membahas penggunaan atau pembelajaran AI dalam konteks sains, matematika, teknik, teknologi, atau STEM terintegrasi. Sintesis tematik menunjukkan empat pola aplikasi utama: sistem tutor dan pembelajaran adaptif, AI generatif untuk penjelasan serta pembuatan kode, analitika dan asesmen berbantuan AI, serta robotika dan aktivitas literasi AI. AI dapat mempercepat umpan balik, mendukung diferensiasi, memperluas eksplorasi ide, dan meningkatkan penyelesaian tugas. Namun, bukti tentang pemahaman konseptual, transfer, kreativitas, dan pemecahan masalah jangka panjang masih tidak merata. Manfaat paling konsisten muncul ketika AI ditempatkan sebagai scaffolding dalam inkuiri, problem-based learning, project-based learning, atau engineering design, disertai verifikasi sumber dan refleksi. Implementasi menghadapi risiko ketergantungan, halusinasi, bias, privasi, integritas akademik, kesenjangan akses, dan keterbatasan kompetensi guru. Kajian menyimpulkan bahwa AI perlu diposisikan sebagai mitra berpikir yang diawasi manusia, bukan pengganti penalaran peserta didik. Advances in artificial intelligence (AI) are expanding how Science, Technology, Engineering, and Mathematics (STEM) learning is designed, enacted, and assessed. This review synthesizes AI applications, pedagogical practices, learning outcomes, and implementation challenges in STEM education. A structured integrative literature review was conducted on 30 core publications from 2019–2025. Literature was identified through scholarly metadata services, education indexes, and publisher websites using combinations of STEM education, artificial intelligence, generative AI, intelligent tutoring system, adaptive learning, learning analytics, educational robotics, and learning outcomes. Publications were included when they examined AI use or AI learning in science, mathematics, engineering, technology, or integrated STEM contexts. The thematic synthesis identified four dominant application patterns: intelligent tutoring and adaptive learning, generative AI for explanation and code production, AI-supported analytics and assessment, and robotics or AI literacy activities. AI can accelerate feedback, support differentiation, broaden idea exploration, and improve task completion. Evidence for durable conceptual understanding, transfer, creativity, and long-term problem-solving remains uneven. Benefits are most credible when AI functions as scaffolding within inquiry, problem-based learning, project-based learning, or engineering design and when learners must verify sources, justify decisions, and reflect on AI contributions. Major challenges include over-reliance, hallucination, bias, privacy, academic integrity, unequal access, and limited teacher capacity. AI should therefore be treated as a human-supervised cognitive partner rather than a substitute for learners’ reasoning.]]>
Copyrights © 2026
