A Questionnaire Study on the Dual-Path Impact of Perceived Usefulness and Self-Efficacy on the Dependence on AI Learning Tools

This study employed a structured questionnaire as the data collection instrument, designed to measure university students' perceived usefulness of AI learning tools, their self-efficacy following tool use, and the resulting dependency behaviors. The questionnaire was developed based on the Technology Acceptance Model and self-efficacy theory, utilizing an anonymous self-report method. All items were presented in Chinese and measured using a five-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree).1. Demographic Variables and Usage BehaviorThe first section of the questionnaire collected participants' background information and their AI tool usage patterns. Specifically, it included: gender, academic year, and disciplinary category. Regarding usage behavior, the questionnaire measured the following: (a) the specific name of the most frequently used AI learning tool (open-ended response); (b) average frequency of use (a 6-level scale from "Multiple times daily" to "Rarely use"); and (c) primary types of learning tasks for which the tool was used (multiple-choice, including six categories: solving difficult problems, generating ideas/outlines for papers/reports, polishing or translating text, programming assistance and debugging, summarizing literature, and simulated conversation practice).2. Measurement of Core VariablesThe main body of the questionnaire consisted of three adapted or self-developed scales corresponding to the study's core variables.· Perceived Usefulness: Measured using a 5-item scale adapted from the classic scale by Davis (1985). This scale assessed the degree to which users believe using AI learning tools enhances their learning performance. A sample item is: "Using AI learning tools can significantly improve my learning efficiency." The internal consistency reliability (Cronbach’s α) for this scale in the present study was 0.901.· Self-Efficacy: Drawing on Bandura's (1977) self-efficacy theory, a 6-item scale comprising two dimensions was constructed. The first dimension, "General Self-Efficacy in Using AI Tools" (3 items), measured students' confidence in operating and controlling AI tools to complete tasks, e.g., "I am confident in skillfully operating my commonly used AI learning tools to complete learning tasks." The second dimension, "AI-Empowered Academic Self-Efficacy" (3 items), measured students' enhanced belief in their overall learning capability due to using AI tools, e.g., "After using AI tools, I feel more confident in tackling more challenging learning tasks (e.g., course papers, complex projects)." The overall internal consistency reliability for this scale was 0.889.· Dependency on AI Learning Tools: To measure dependency behavior, a self-developed 7-item scale covering behavioral and affective-cognitive aspects was used. The "Behavioral Dependency" sub-dimension (3 items) focused on the priority, indispensability, and increasing trend of tool use, e.g., "When starting a learning task, my first thought is to use an AI learning tool." The "Affective and Cognitive Dependency" sub-dimension (4 items) involved emotional reactions to tool outputs, trust tendencies, and concerns about over-reliance, e.g., "I feel anxious or disappointed when the answers provided by the AI tool do not match my expectations." The overall internal consistency reliability for this scale was 0.876.3. Data Quality Control and Supplementary InformationAll scales were pilot-tested prior to the formal survey to ensure item clarity and content validity. Two open-ended questions were included at the end of the questionnaire. These aimed to qualitatively understand the specific scenarios where AI tools provided the greatest help, as well as the situations and reasons why users intentionally reduce or avoid their use. This provides contextual, in-depth supplementary information for interpreting the quantitative results.This detailed description of the questionnaire ensures transparency regarding the research measurement instruments, provides a solid foundation for data analysis and interpretation, and supports the reproducibility of the study.