The Impact of AI-Powered Text Generation Tools on the Critical Thinking Skills of Undergraduate Students

Abstract

The swift adoption of generative AI (GAI) technologies such as ChatGPT in higher education has raised important questions about their effect on students' critical thinking—an essential component of undergraduate education. This study adopts a mixed-methods approach to investigate the influence of AI-based text generation tools on the enhancement of cognitive skills among 100 undergraduate students, using Bloom’s Taxonomy as a framework. Employing a quasi-experimental design, participants were divided into a control group and an experimental group. Over a span of six weeks, the experimental group engaged with ChatGPT to complete a series of structured academic exercises. Pre- and post-test assessments, alongside thematic analysis of student surveys, revealed dual outcomes: while ChatGPT significantly enhanced foundational skills (remembering, understanding, applying), it showed negligible effects on higher-order critical thinking (analyzing, evaluating, creating). Students reported using AI for proofreading, structuring assignments, and literature summarization, freeing cognitive bandwidth for Higher tasks. However, challenges such as AI "hallucinations" (fabricated outputs), digital literacy gaps, and cultural biases underscored risks of over-reliance and superficial engagement. Qualitative findings highlighted ethical dilemmas, including concerns about diminished empathy and academic integrity, while also noting AI’s role in fostering metacognitive reflection through fact-checking. The study advocates for a balanced pedagogical approach, where AI complements—rather than replaces—human-guided instruction, and proposes institutional strategies for AI literacy training, ethical guidelines, and curriculum redesign. These findings contribute to the discourse on technology-enhanced learning, emphasizing the need to preserve cognitive rigor in an AI-driven educational landscape.