Abstract

Generative AI tools are rapidly reshaping how undergraduate students seek and engage with academic information, yet empirical understanding of what drives adoption and how students behave when using these tools relative to traditional search engines remains limited. This study examined undergraduate students’ information-retrieval behaviors and perceptions when using ChatGPT compared to Google Search, using the technology acceptance model (TAM; Davis, 1989) extended to include trust (Gefen et al., 2003) as the guiding framework. An embedded mixed-methods design was employed, in which Likert-scale perception ratings provided a structured quantitative signal and thematic analysis of semi-structured interviews, structured information-retrieval tasks, and concurrent think-aloud protocols served as the dominant interpretive strand. Ten undergraduate students at a large private university in the western United States completed two counterbalanced sessions, one with each tool. Findings suggest that trust functions less as a coequal TAM construct and more as a gating variable that conditions whether perceived ease of use and perceived usefulness translate into realized advantage. Source transparency emerged as the primary mechanism through which students regulated trust in ChatGPT. Its absence eroded confidence, while its presence restored it. Students calibrated trust thresholds to task stakes and navigated the two tools through structurally different trust-resolution pathways: a layered hierarchy of behaviors on Google and a single-prompting lever on ChatGPT. Notably, ChatGPT adoption continued despite an incomplete trust foundation, a pattern interpreted through Sunstein’s (2026) “product trap” framework, in which sustained use is driven less by individual confidence than by external dependencies and competitive pressure. Implications for AI literacy instruction and institutional response are discussed.

Degree

MS

College and Department

David O. McKay School of Education; Instructional Psychology and Technology

Rights

https://lib.byu.edu/about/copyright/

Date Submitted

2026-06-20

Document Type

Thesis

Keywords

technology acceptance model, generative artificial intelligence, information retrieval, trust in ai, large language models, technology adoption, information landscape, prompt engineering, AI literacy

Language

english

Download

Included in

Education Commons

Share

COinS