Student Concerns and Perceived Negative Cognitive Effects of AI Tool Use in Academic Settings: A Text Mining Analysis
Main Article Content
Abstract
The growing use of artificial intelligence (AI) tools in academic work has raised concerns about students’ dependence on these tools and possible changes in their learning behaviors. This study examined perceived negative cognitive and behavioral changes reported by 197 college students in relation to their regular academic use of AI tools. Their open-ended responses were analyzed using word-frequency analysis, Latent Dirichlet Allocation (LDA), Non-negative Matrix Factorization (NMF), and TextBlob sentiment analysis. Topic solutions from two to ten topics were compared based on coherence, consistency across repeated LDA runs, interpretability, and agreement between LDA and NMF. Two topics were retained: behavioral disengagement and AI dependency, and reduced independent cognitive and academic engagement. LDA assigned 130 participants (66%) to the first topic and 67 (34%) to the second, while NMF assigned 124 (62.9%) and 73 (37.1%), respectively. The two models assigned 177 participants (89.8%) to corresponding topics, with κ = .779. The mean TextBlob sentiment score was .090 (SD = .169). TextBlob scores did not differ significantly by sex, U = 4055, p = .081, and showed a small positive relationship with response length, rₛ = .266, p < .001. The findings show that students commonly described dependence, reduced effort, and concerns about independent thinking.