Improving Higher Order Thinking Skills through Project-Based Learning with AI-Enhanced Formative Assessment

Abstract

The low level of students’ Higher Order Thinking Skills (HOTS) in computer science presents a challenge in the implementation of the Merdeka Curriculum. This study examines the extent to which a Project-Based Learning (PBL) approach combined with AI-enhanced formative assessment can improve students’ HOTS in computer networking, particularly their ability to analyze computer network concepts and apply them in real-life situations. This study contributes by presenting an innovative learning model that integrates PBL and AI-enhanced formative assessment as a practical alternative to improve HOTS in Indonesian high schools. The study used a classroom action research (CAR) design as the overarching framework, implemented as a single CAR cycle of planning, action, observation, and reflection; within this cycle, a mixed-methods approach combining a quasi-experimental one-group pretest-posttest method with a descriptive qualitative method was used to collect and analyze data, and the study involved 33 eleventh-grade students at SMA N 3 Purwakarta. The intervention integrated a digital poster project with layered formative assessment using ChatGPT and Padlet. Data were analyzed via paired t-tests and thematic analysis using NVivo. Results showed a significant cognitive improvement, with the average score rising from 74.55 to 83.64 (p = 0.026). Qualitatively, 58% of students were able to evaluate connectivity options based on real-world needs, and each group conducted an average of 3.2 rounds of revisions based on AI feedback. It was concluded that the integration of PBL with AI-enhanced formative assessment effectively fosters students’ higher-order thinking skills (HOTS) at the school studied, though the single-school sample means the findings are best generalized at the regional rather than national level.