Abstract:

In recent years, with its application in teaching and learning becoming increasingly prevalent, the rapid advancement of AI has profoundly impacted the education field. Traditionally, students' knowledge construction, critical thinking, and creativity were fostered through extensive reading, writing, and discussion. However, with AI now substituting portion of these activities, concerns arise about the potential negative effects on students' knowledge construction.

This research paper is driven by the need to investigate both the positive and negative impacts of AI on knowledge construction and to develop a balanced teaching framework. The primary goal is to ensure that AI enhances, rather than diminishes, students' critical thinking and creativity. Grounded in epistemology as the theoretical foundation, this study examined whether AI applications in education promote the development of students' critical thinking and creativity.

This mixed-methods study involved 27 graduate students using "Sharestart" and "Visible Thinking" pedagogies on Miro to compare non-AI versus AI-assisted discussion phases. Data collected via surveys, interviews, and artifacts were analyzed using MANOVA and ANOVA quantitatively, while qualitative analysis employed Content Analysis and AI model evaluation.

In the quantitative analyses, MANOVA results indicated significant AI effects (Wilks’ Λ=.544, F(5,83)=13.937, p<.001). Subsequent two-way ANOVAs confirmed AI usage significantly enhanced Technical (F=21.997, p<.001), Learning (F=40.303, p<.001), and Critical Thinking (F=5.335, p=.023) factors, while time and interaction effects remained non-significant.

In the qualitative analysis utilizing the trained Gemini 3.0 Pro model to evaluate 194 AI-assisted and 63 non-AI thinking screenshots, AI outputs demonstrated superior depth in Technical, Learning, and Critical Thinking factors. Conversely, non-AI human outputs predominantly emphasized Social and Emotional factors, focusing on classroom dynamics and interpersonal connections.

In the semi-structured interviews of 12 participants, verbatim transcripts were coded using directed content analysis. Results indicated that Technical (55) and Learning (28) factors dominated when using AI, positioning it as a cognitive scaffold. Conversely, Social (33) and Affective (32) factors prevailed without AI, highlighting human-centric concerns.

The findings confirm that AI integration significantly enhances knowledge construction, specifically in Technical, Learning, and Critical Thinking domains, countering concerns that AI diminishes intellectual engagement. Quantitative and qualitative results consistently position AI as a powerful cognitive scaffold for analytical rigor. However, the qualitative analysis reveals a critical divergence that non-AI human effort remains superior in fostering Social and Affective connections. Therefore, to achieve the proposed balanced teaching framework, educators should leverage AI for structural depth while intentionally preserving human-centric interactions to maintain classroom dynamics and emotional resonance.

Keywords:

Artificial Intelligence, Creativity, Critical Thinking, Epistemology.