Agentic education · Learning science

Actionable companion · Feedback & scaffolding

Learning with agency: Teacher or Artificial Intelligence? The Effect of Decision-Making Agent on Junior High School Students’ Decision Acceptance: A Moderated Mediation Model

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): Secondary school. Verify the participant population and task before transferring the findings.

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

Artificial intelligence (AI) is increasingly used in educational decision-making. This raises important questions about how students perceive and accept outcomes from non-human agents. Based on authority acceptance theory and organizational justice theory, this study explored three research questions: how decision-making agents (teachers vs. AI) affect junior high school students’ acceptance of educational decisions, the mediating mechanism of perceived fairness, and the moderating effect of explanatory feedback. This study adopted a 2 (decision-making agent: teacher vs. AI) × 2 (explanatory feedback: with vs. without explanation) between-subjects experimental design, with 250 seventh-grade students as participants. Results showed that students reported significantly higher decision acceptance and perceived fairness for teacher-made decisions than for AI-generated decisions. Perceived fairness partially mediated the relationship between the decision-making agent and decision acceptance. The mediating effect accounted for 67.52% of the total effect. Explanatory feedback positively moderated only the path from the decision-making agent to perceived fairness, with no significant moderating effect on the direct link between the decision-making agent and decision acceptance. Provision of explanations widened the human–AI gap in perceived fairness, which in turn further enlarged the gap in decision acceptance via the mediating pathway of perceived fairness. These findings extend research on explanatory feedback to the basic education context, while highlighting the significance of teacher-led human–AI collaboration and interpretable AI design amid the digital transformation of education.

What kind of evidence is this?

Supporting AI learning science · Survey / observational study

Stage: Secondary school. 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

Before formal hypothesis testing, GPower 3.1 was used for a priori power analysis to calculate the minimum sample size for the four-group between-subjects ANOVA (main effects and interaction effects). Consistent with existing educational research on decision acceptance and fairness perception, we set the medium effect size f = 0.25, α = 0.05, target statistical power 1 − β = 0.85, numerator df = 1, and four experimental groups. The simulation output noncentrality parameter λ = 9.12, critical F = 3.90, denominator df = 142, and the minimum valid sample size N = 146 to effectively detect the hypothesized main and interaction effects. Sample size was calculated in advance via GPower 3.1, which indicated a minimum sample of 146 participants to achieve a statistical power of 0.85, and a total of 258 seventh-grade students were recruited from a junior high school in a city of Henan Province, China, notably, statistical power varies substantially across different pathways within the full moderated mediation model, as a sample size sufficient to identify simple main factor effects may fail to provide adequate power to detect weaker interaction terms or the index of moderated mediation given that conditional indirect and interactive effects generally produce smaller population effect sizes than main effects, and our final valid analytical sample of 250 participants retained after data screening far exceeds the power-derived minimum threshold of N = 146, which guarantees satisfactory statistical power to test all core hypotheses covering group main effects, two-way interaction effects, and the moderated mediation index in the structural model.

Limitations and uncertainty · source passage

This study preliminarily uncovers how a decision-making agent shapes students’ acceptance of decisions in educational settings, yet several limitations exist. First, participants were only grade seven junior high students from one city in Henan Province. The narrow sample limits representativeness, so findings should be generalized cautiously to students from other regions or schooling stages. Second, this research relies on scenario-based experiments. Such simulated settings differ from real-class homework grading and on-site feedback, lowering the ecological validity of the results. Third, only standardized math homework scoring is examined. Subjective and complicated educational decisions, including mental health assessment, reward and punishment management, and comprehensive quality evaluation, are excluded, restricting the generalizability of conclusions. Finally, we must admit that in the experiment, we set the teacher as the decision-making agent for the students, who were more familiar with the teacher. This might have led to the confounding of multiple interrelated variables, such as the long-term familiarity between the teacher and the students, stable interpersonal trust, the teacher’s accountability status within the school, local classroom authority, and the teacher’s exclusive subject-specific contextual knowledge. Therefore, although this design can compare the two types of real decision-making agents in the students’ school environment, it has good ecological validity, but it cannot separate the pure effect of “whether the decision-maker is a human or an AI” from the accompanying familiarity, relationship trust, and authority perception.

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.

  • Give feedback on the learner’s reasoning, then fade hints. Record whether the learner can complete the next task without that scaffold.

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. Ask learners to attempt a problem unaided, compare an agent hint, explain a revision and verify the final reasoning against a trusted source. Make the permitted use explicit for each assignment. 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. Measure unaided transfer and delayed retention, calibration of confidence and the ability to identify an incorrect AI explanation. Record whether help-seeking becomes answer copying. 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

Zhen Zhang; Siyu Fan; Chunhui Qi. Teacher or Artificial Intelligence? The Effect of Decision-Making Agent on Junior High School Students’ Decision Acceptance: A Moderated Mediation Model. Behavioral Sciences (2026-07-19). https://doi.org/10.3390/bs16071227

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.