CADGPT Framework: Model Project-Based Learning Berbantuan AutoCAD dan ChatGPT untuk Meningkatkan Hasil Belajar pada Mata Kuliah Konstruksi Bangunan Gedung

Abstract

The digital transformation of the Architecture, Engineering, and Construction (AEC) industry requires learning approaches that integrate pedagogical strategies, professional software, and artificial intelligence. This study aims to examine differences in students' cognitive learning outcomes between the implementation of the CADGPT Framework, which integrates Project-Based Learning (PjBL), AutoCAD, and ChatGPT as an AI Learning Assistant, and conventional instruction in the Building Construction course. A quasi-experimental design employing a nonequivalent pretest–posttest control group was used, involving 60 students divided into an experimental group and a control group, with 30 students in each. Participants were selected using intact group sampling, involving two existing classes without individual randomization. The study was conducted over eight sessions, including the pretest and posttest, and learning outcomes were assessed using a 30-item multiple-choice test. Data were analyzed using descriptive statistics, t-tests, Normalized Gain (N-Gain), the Mann–Whitney U test, and Cohen's d. The results showed that the experimental group achieved a higher mean posttest score (79.33) than the control group (72.78), with a statistically significant difference (t(58) = 2.701, p = 0.009). The reported N-Gain was also higher in the experimental group (0.6257) than in the control group (0.4760), with a significant difference in N-Gain distributions or ranks (U = 188.000, p < 0.001). Cohen's d of 0.70 indicated a medium effect size. These findings suggest that the implementation of the CADGPT Framework was associated with higher cognitive learning outcomes in the two classes examined. However, the individual contributions of PjBL, AutoCAD, and ChatGPT could not be isolated because these components were implemented as an integrated instructional approach.