Actionable companion · Self-regulation & metacognition
Learning with agency: The personalization paradox: how AI-driven adaptive learning environments are associated with college students’ academic emotions and self-regulation learning—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.
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:
Background Artificial intelligence (AI)-engaged adaptive studying platforms are being rapidly adopted across higher education. While they provide significant opportunities to tailor students with both educational content and learning paces to individual needs, educators and researchers proposed their concerns that over-reliance on algorithmic itself may unconsciously be negatively associated with students’ ability in self-regulated learning (SRL)—a phenomenon termed the “personalization paradox.” Aims Integrating control-value theory (CVT) and self-regulated learning (SRL) theory, this study demonstrates that academic emotions—enjoyment, anxiety, and boredom—mediate the relationship between perceived AI-adaptive environments and SRL, with CVT clarifying why these environments foster emotional responses and SRL theory explaining how such emotions translate into behavioral outcomes. This study uniquely combines CVT and SRL within a single moderated mediation structure, providing an innovative theoretical lens to know the personalization paradox very well. Methods A three-wave longitudinal survey was applied across one semester during the academic year with 486 undergraduates from four universities in China. At T1, students reported their perceptions of the AI-adaptive surroundings and AI literacy to researchers; at T2, they reported academic emotions; at T3, they reported their SRL behaviors. Structural equation modeling with latent interaction terms and cluster-robust standard errors were utilized to estimate the moderated mediation structure. Results Perceived AI-adaptive environments were negatively related to SRL (β = −0.19, p < 0.01), giving certain supports to personalization paradox. Indirect effects in enjoyment (β = −0.08), anxiety (β = −0.06), and boredom (β = −0.09) were completely significant, respectively accounting for 54.8% of the entire effect. Most importantly, AI literacy moderated the first selective pathways from the AI environment to every academic emotion (e.g., enjoyment was β = 0.14, p < 0.01). such that the negative emotional impacts were moderated among students with higher AI literacy. This model accounted for 28.4% of the variance in SRL. Conclusion The personalization paradox operates through affective pathways, with AI literacy playing the role as a significant protective index. These results highlight the emotional and autonomy-related trade-offs inherent in algorithmic personalization and suggest that universities embed AI literacy as a credit-bearing module and design adaptive systems with lucid, user-controllable recommended interfaces. Causal inferences still are limited given the existing observational design.
What kind of evidence is this?
Intelligent tutoring & adaptive learning · Survey / observational study
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
To capture the dynamic yet progressive nature of autonomy erosion—a whole process that steps forward progressively over time—we employed a three-wave longitudinal research design, which overcomes the limitations of cross-sectional designs since they are insufficient to achieve our objectives (Little, 2013). Data collection occurred from September 2025 to January 2026. The three waves scheduled at the beginning of the semester (Weeks 2–3, T1), the mid-semester period (Weeks 9–10, T2), and the end of the semester (Weeks 16–17, T3), as shown in Figure 2.
Limitations and uncertainty · source passage
Several limitations warrant acknowledgement. First, the observational design, while incorporating temporal separation between measurement waves, precludes final causal inference. In the future, the similar research should employ randomized controlled trials or cross-lagged panel designs with baseline controls to establish causality. Second, reliance on self-reported measures introduces potential biases, future studies would benefit from integrating behavioral logs, eye-tracking data, or physiological indices. Third, the sample was restricted to Chinese university students limiting cross-cultural generalizability, particularly given that AI acceptance and trust may vary across cultural contexts. Fourth, the aggregation of diverse AI tools into a single composite measure may mask tool-specific effects; future research should implement multilevel analyses that distinguish among tool specifics. Fifth, common method bias, while assessed and found minimal, cannot be totally ruled out; multi-method designs could strengthen future researches.
Design the system before the pilot
Document the learner model, mastery rule, hint sequence and adaptation signal. Compare adaptive routing with the same content in a non-adaptive sequence. Check whether early errors lead to persistent low-expectation pathways.
- 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
- 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.
- 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.
- 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.
- 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
Jing Li; Ziheng Lin; Chengqun Qiu. The personalization paradox: how AI-driven adaptive learning environments are associated with college students’ academic emotions and self-regulation learning—a moderated mediation model. Frontiers in Psychology (2026-09-11). https://doi.org/10.3389/fpsyg.2026.1915839
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.