Does ChatGPT-enhanced collaborative learning foster critical thinking in education? A Bloom’s Taxonomy perspective

• ChatGPT-based collaborative learning boosts comprehension, applying, and memory. • Understanding drives critical thinking, while applying plays a moderate role. • NCA shows understanding and applying are vital prerequisites for critical thinking. • fsQCA identifies multiple paths to high or low critical thinking outcomes. • AI alone cannot ensure critical thinking; structured, interactive learning is key. This study examines how ChatGPT-supported collaborative learning influences critical thinking in health education using Bloom’s Taxonomy. Purposive sampling was used to collect data from 665 Indonesian health students through an online survey. Partial Least Square – Structural Equation Modelling (PLS-SEM) assessed the direct effects of cognitive processes on critical thinking. Necessary Condition Analysis (NCA) identified essential cognitive conditions, while fuzzy sets qualitative comparative analysis (fsQCA) explored different cognitive pathways leading to high or low critical thinking. Collaborative learning significantly enhances understanding, applying, and remembering. Understanding has the strongest effect on critical thinking, while applying and remembering have moderate effects. These findings suggest that deep comprehension drives analytical reasoning, whereas applying and remembering serve complementary roles. NCA confirms that understanding and applying are necessary for fostering critical thinking, while remembering plays a supporting role. fsQCA results indicate that students who combine deep understanding with memory retention exhibit strong critical thinking. In contrast, students who rely solely on remembering without comprehension or application struggle to develop higher-order reasoning. This study reveals that ChatGPT does not inherently enhance critical thinking but must be integrated into structured collaborative learning. Effective AI-assisted education requires active discussion, application, and critical evaluation of AI-generated insights. These findings offer a framework for optimizing AI-driven health education to support both knowledge acquisition and analytical reasoning in clinical decision-making.