Design and preliminary validation of an AI enhanced game based learning environment for computational thinking in upper secondary education

Abstract Integrating generative artificial intelligence within game-based learning offers promising but still-maturing possibilities for computational thinking (CT) education. This developmental research study (Type I) reports the design and preliminary validation of an AI-enhanced game-based learning environment (AI-GBLE) for upper-secondary students. Grounded in constructivism, cognitive load theory, information processing theory, mental models theory, and game-based learning principles, the environment integrates six components on the Roblox platform—problem-based scenarios, learning resources, collaborative learning, a CT enhancement centre, scaffolding, and an AI coaching system combining a GPT-4-powered conversational non-player character with a rule-based adaptive difficulty engine—within an authentic Smart Greenhouse (Internet of Things) context. Following expert appraisal of content, instructional design, and media-technology quality by nine specialists, three per domain, 30 Grade 11 students in Thailand used the environment across eight 60-minute sessions over six weeks in a one-group pretest–posttest validation. Students showed pre–post gains of 3.07 points on the 20-point computational thinking test (Wilcoxon r = .86; Cohen’s d_av = 2.79) and 5.90 points on the 15-point achievement test (d_av = 3.23)—standardised effects inflated by ceiling compression—and reported strong immediate satisfaction across six acceptance dimensions (overall M = 4.64). Because the single-group design precludes causal attribution, these findings are interpreted as preliminary evidence—of feasibility, acceptability, and learning-related signals from a developed prototype—rather than as intervention effectiveness. The individual contributions of the AI coach, adaptive engine, and other components were not empirically established. The study contributes a theory-mapped design heuristic for AI-augmented game-based learning and identifies controlled comparisons, improved measurement, and mechanism-focused analyses as priorities for subsequent research.