Agentic education · Learning science

Actionable companion · Self-regulation & metacognition

Learning with agency: Between Human and Machine: University Teachers’ Accounts of Judgment, Metacognitive Defense, and Cognitive Offloading in Generative AI Use

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:

Generative artificial intelligence (GenAI) allows university teachers to delegate cognitive work while deciding how students should engage with it. Using codebook thematic analysis, we examined multi-source accounts from 25 teachers recruited through an AI-focused learning community and its professional networks in China. Materials comprised interviews with 22 teachers, three documentary cases, self-narratives and reflective essays. Three themes organize the findings: reconstructing the teacher role, defending professional judgment, and managing teachers’ and students’ cognitive participation. Participants described disciplinary expertise as a basis for evaluating AI outputs and reported practices intended to retain human engagement. Instrumental and partnership orientations overlapped; descriptive comparison with later use profiles did not establish a simple relationship between partnership language and usage intensity. Because the protocol introduced several focal topics, we distinguish prompted positions from volunteered formulations. The proposed belief–strategy–practice cycle is a heuristic interpretation of retrospective accounts, not a demonstrated mechanism. The study contributes an account of how teachers connect their own cognitive offloading with their reported organization of students’ learning opportunities. Its claims concern this AI-engaged sample rather than university teachers generally.

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

We adopted a basic qualitative design (Merriam, 2009), comparing across cases to examine both commonality and variation in teachers’ accounts. The analytic corpus comprises 25 teacher cases (Table 2). Of the 25, 22 took part in semi-structured interviews; for the remaining three (Teachers D, E, and F), no interview was conducted, and their case materials consist of field notes together with, respectively, chat records, product documents, and published essays. The first author’s reflective texts are not part of the coded corpus and contribute no evidence to the analytic case counts, coverage figures or stance distributions; they are used as reflexive material and are described in Section 3.5.

Limitations and uncertainty · source passage

Drawing on multi-source qualitative materials from 25 university teachers recruited through an AI-focused learning community and its professional networks, this study examined how they describe their beliefs about GenAI and the cognitive strategies they use alongside it. Four conclusions follow. First, participants described a broadly shared boundary around what AI cannot do—emotion, embodiment, metacognition, tacit knowledge, judgment—which served both to make sense of the technology and to protect a professional position; this boundary was elicited by a direct question and should be taken as the shape of their answers to it. Second, participants commonly reconstructed the teacher role from “knowledge transmitter” towards “guide, designer, companion, gatekeeper”, in terms they volunteered, while ten of the 25 cases (40.0%) retained a reinterpreted claim to expertise. Third, judgment and gatekeeping were re-centered as the defended core of professional work, with verification and judgment carrying clear evidence in 24 of 25 cases. Fourth, participants described managing cognitive participation on both sides of the desk, taking divergent positions on “intellectual degeneration” (9 endorsement, 11 conditional, 2 rebuttal, 1 ambivalent, 2 no position) while describing a varied repertoire of defensive strategies with differing levels of explicit support. The relation between belief and cognition that these accounts suggest is reciprocal but not demonstrated: it is an interpretation of what teachers say, offered for longitudinal and observational examination.

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 plan–attempt–check–reflect prompts and compare confidence with actual performance. Keep learner choice and reflection visible.
  • Measure engagement separately from achievement. Check whether novelty-driven participation persists and whether it translates to independent performance.

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

Zhen Qiang; Zhuo Wang. Between Human and Machine: University Teachers’ Accounts of Judgment, Metacognitive Defense, and Cognitive Offloading in Generative AI Use. Journal of Intelligence (2026-10-10). https://doi.org/10.3390/jintelligence14100254

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.