<![CDATA[Kreativitas siswa dalam merancang antarmuka digital dapat terhambat ketika pembelajaran terlalu bergantung pada template siap pakai. Penelitian ini bertujuan membandingkan kreativitas desain antarmuka siswa yang belajar dengan fitur AI Text-to-Image pada Canva dan siswa yang menggunakan fitur standar Canva, mengukur peningkatan kreativitas, serta menjelaskan pengalaman siswa selama ko-kreasi dengan AI. Penelitian menggunakan metode campuran dengan desain sekuensial eksplanatori. Tahap kuantitatif menerapkan kuasi-eksperimen nonequivalent control group pada 61 siswa kelas VIII SMP Negeri 2 Sijunjung, terdiri atas 30 siswa kelompok eksperimen dan 31 siswa kelompok kontrol. Kreativitas dinilai melalui tes unjuk kerja berdasarkan kelancaran, keluwesan, orisinalitas, dan elaborasi; tahap kualitatif menggunakan wawancara reflektif terhadap enam siswa kelompok eksperimen. Analisis tanpa penyesuaian skor awal menunjukkan rata-rata posttest kelompok eksperimen (12,80) lebih tinggi daripada kelompok kontrol (9,20), t(59) = 8,712, p < 0,001, Cohen’s d = 2,23, 95% CI [1,58, 2,87]. Secara deskriptif, N-Gain kelompok eksperimen sebesar 0,66 dan kelompok kontrol 0,29, dengan selisih indikator terbesar pada orisinalitas. Wawancara menunjukkan bahwa siswa terbantu memunculkan ide visual, merevisi prompt secara iteratif, dan merefleksikan tanggung jawab etis. Temuan ini menunjukkan hasil yang menjanjikan bagi integrasi AI Text-to-Image melalui prompt engineering terstruktur, tetapi interpretasi kausal perlu dibatasi karena penugasan kelas tidak dilakukan secara acak. Students’ creativity in digital interface design can be constrained when learning relies heavily on ready-made templates. This study compared interface-design creativity between students learning with Canva’s AI Text-to-Image feature and those using standard Canva features, quantified improvement, and explored students’ experiences during human–AI co-creation. An explanatory sequential mixed-methods design was employed. The quantitative phase used a nonequivalent control group quasi-experiment involving 61 eighth-grade students at SMP Negeri 2 Sijunjung, with 30 students in the experimental group and 31 in the control group. Creativity was assessed through a performance task covering fluency, flexibility, originality, and elaboration; the qualitative phase used reflective interviews with six experimental-group students. The unadjusted posttest comparison showed a higher mean for the experimental group (12.80) than the control group (9.20), t(59) = 8.712, p < .001, Cohen’s d = 2.23, 95% CI [1.58, 2.87]. Descriptively, normalized gain was 0.66 in the experimental group and 0.29 in the control group, with the largest indicator-level difference in originality. Interviews indicated that students generated visual ideas more easily, revised prompts iteratively, and reflected on ethical responsibility. These findings suggest promising educational value for structured prompt-engineering activities with text-to-image AI, while causal claims should remain cautious because class assignment was nonrandom. ]]>
Copyrights © 2026
