Analisis miskonsepsi siswa berbasis artificial intelligence pada materi suhu dan kalor

Abstract

Abstract

This study aimed to analyze the level and patterns of misconceptions among eleventh-grade students on the topic of temperature and heat, involving 200 respondents from SMA Negeri 1 Percut Sei Tuan, MAN 2 Model Medan, and SMA PAB 4 Sampali, as well as to examine the effectiveness of an Artificial Intelligence (AI)-based diagnostic assessment. Misconceptions often arise from students’ everyday experiences that are not scientifically grounded, while manual identification of misconceptions tends to be time-consuming and inefficient. The study employed a mixed-methods approach with a sequential exploratory design. The instruments used included teacher interviews, an AI-based two-tier diagnostic test, and student and teacher response questionnaires. Data analysis involved quantitative descriptive analysis to determine the percentage of students’ conceptual understanding and qualitative analysis to identify misconception patterns based on combinations of answers and reasoning. The findings revealed that 31% of students demonstrated proper conceptual understanding, 46% experienced dominant misconceptions, and 23% lacked conceptual understanding. The dominant misconceptions were related to the concepts of temperature and heat, heat transfer, thermal equilibrium, and phase changes. The AI-based diagnostic assessment proved effective in rapidly identifying misconceptions and providing consistent feedback. In addition, student responses were categorized as very good (74.99%), while teacher responses were categorized as very good (81.94%). Therefore, AI-based diagnostic assessment has strong potential to improve adaptive, data-driven physics evaluation practices.

Keywords: misconceptions; artificial intelligence; two-tier diagnostic test; temperature and heat.

Abstrak

Penelitian ini bertujuan menganalisis tingkat dan pola miskonsepsi siswa kelas XI dengan responden sebanyak 200 dari SMA Negeri 1 Percut Sei Tuan, MAN 2 Model Medan, dan SMA PAB 4 Sampali pada materi suhu dan kalor, serta mengkaji efektivitas asesmen diagnostik berbasis Artificial Intelligence (AI); miskonsepsi sering muncul dari pengalaman sehari-hari tanpa landasan ilmiah dan identifikasi manual memakan waktu lama. Menggunakan pendekatan mixed-methods desain sequential exploratory dengan instrumen wawancara guru, tes diagnostik dua tingkat (two-tier) berbasis AI, serta angket respon siswa-guru, analisis data mencakup deskriptif kuantitatif untuk persentase pemahaman konsep dan kualitatif untuk pola miskonsepsi berdasarkan kombinasi jawaban-alasan; hasil menunjukkan 31% paham konsep, 46% miskonsepsi dominan (pola utama: suhu-kalor, perpindahan panas, keseimbangan termal, perubahan fase), 23% tidak paham, dengan AI efektif identifikasi cepat dan umpan balik konsisten; respon siswa sangat baik (74,99%) dan guru sangat baik (81,94%), sehingga asesmen AI berpotensi tingkatkan evaluasi fisika adaptif berbasis data.

Kata kunci: miskonsepsi; artificial intelligence; tes diagnostik two-tier; suhu dan kalor.