Agentic education · Learning science

Actionable companion · Assessment & reasoning

Learning with agency: Development and validation of the AI dependence scale for Chinese undergraduates and a preliminary exploration

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not claims or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paperSearch paper agentCC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Higher education. Verify the participant population and task before transferring the findings.

Original authors’ abstract, reproduced under the linked source license:

Introduction With the proliferation of generative artificial intelligence (AI) in higher education, student overreliance has become a growing concern, potentially undermining critical thinking and autonomous learning. To address the lack of a comprehensive measurement tool, this study developed and validated the AI Dependence Scale (AIDep-22), a new instrument designed to assess this phenomenon across four hypothesized dimensions: emotional dependence, functional dependence, cognitive dependence, and loss of control. Methods The scale was constructed following a rigorous two-stage process, beginning with item generation and refinement through expert reviews and cognitive interviews, followed by psychometric evaluation with two independent samples of Chinese university students (N = 400 each). Results An exploratory factor analysis (EFA) supported the four-factor structure, which was subsequently confirmed by a confirmatory factor analysis (CFA) on the second sample. The final 22-item scale demonstrated excellent internal consistency (Cronbach's alpha = 0.87), strong convergent and discriminant validity, and robust criterion-related validity. Preliminary analyses also identified key demographic risk factors, revealing that male students, upper-year students, those in applied majors, and more frequent AI users reported significantly higher dependence. Discussion This study contributes a reliable and valid diagnostic tool that enables educators and researchers to identify and support students at risk, and to design targeted interventions that promote a more balanced human-AI relationship in higher education.

What kind of evidence is this?

Supporting AI learning science · Qualitative / mixed methods

Stage: Higher education. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Study methods · source passage

The development and validation of the AIDep-22 adhered to the classical paradigm for scale construction articulated by (Hinkin 1998) and (DeVellis and Thorpe 2021). The research was structured in two main stages. The initial stage involved three foundational phases: (1) item generation, (2) content validation, and (3) pilot testing. Following this, a second stage was conducted, which consisted of a large-sample psychometric evaluation. This evaluation was designed to establish the scale's factor structure through exploratory and confirmatory factor analyses (EFA and CFA), assess its internal consistency, and confirm its convergent and discriminant validity. Furthermore, the scale's criterion validity was examined by assessing its associations with academic self-efficacy and AI usage behaviors. As a final step, an initial application of the instrument involved an exploratory analysis of AI dependence levels across different student demographic groups.

Limitations and uncertainty · source passage

While the present study provides robust evidence for the reliability and validity of the AIDep-22, several limitations should be acknowledged. First, the data were collected from a single university in Southwest China, a limitation that may restrict the generalizability of the results to other institutional or cultural contexts. Future research should replicate this study into multi-site and cross-cultural samples to test the stability of the factor structure and demographic patterns across diverse educational systems. Second, the study relied on self-reported measures, which are inherently subject to social desirability and recall biases. Future work should incorporate objective behavioral data, such as log data from AI platforms, to triangulate the self-reported dependence scores more objectively. Third, the study's cross-sectional design precludes causal inferences. Consequently, longitudinal designs are needed to examine the reciprocal dynamics between AI usage and the development of dependence over time.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

  • Use common rubrics, blind scoring where feasible and a fresh unaided reasoning task. Document permissible assistance and avoid treating AI-detector scores as proof of misconduct.

A 90-day district pilot

  1. Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
  2. Days 16–30 · Design. Choose one disciplinary task. Require a first attempt, a documented AI consultation, a critique of the output and an independent explanation. Align the comparison group and scoring rubric with the same objective. Name an accountable owner and set the baseline, success threshold and budget before implementation.
  3. Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
  4. Days 61–90 · Decide. Blind-score reasoning and accuracy on a new task without AI. Distinguish model examination scores, student perceptions and actual learner gains; only the last directly tests educational benefit. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Houyu Wu; Haiyang Ni; Wenfu Luo; Tenglong Wu. Development and validation of the AI dependence scale for Chinese undergraduates and a preliminary exploration. Frontiers in Psychology (2026-01-19). https://doi.org/10.3389/fpsyg.2025.1725393

Companion prepared 2026-10-30. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.