Professional Learning Communities are widely recognized as effective structures for fostering collaborative and situated teacher learning. Their foundational principles—shared vision, collective inquiry, and deprivatized practice—have been consistently linked to improvements in instructor efficacy and student outcomes [1]. Recent scholarly work has investigated virtual PLCs, acknowledging their potential to overcome spatial barriers while also noting challenges related to cultivating trust and maintaining sustained engagement in purely online settings [2].

Contemporary scholarship has focused on the capacity of AI to deliver personalized feedback, such as through automated writing evaluation and classroom discourse analysis [3]. Research also explores AI’s role in curating instructional resources and prompting reflective practice [4]. However, the prevailing literature primarily positions AI as a tool for individual instructor support or direct student instruction. Its systematic integration as a mediating agent within a collaborative, community-based TPD model remains insufficiently explored, representing a notable gap in the field.

Academic inquiry into education in Xinjiang emphasizes the complexities of bilingual and multilingual instruction alongside pronounced regional disparities [5]. Studies advocate for context-sensitive methodologies that respect cultural diversity while promoting proficiency in the national lingua franca. Although technology-assisted learning initiatives have been introduced, there is a scarcity of empirical research examining AI-facilitated, community-oriented TPD models explicitly designed for this distinctive context [5].

Existing research exhibits three principal limitations: i) a disconnect between studies on collaborative PLCs and those focusing on individualized AI applications; ii) a paucity of empirical, mixed-methods investigations into integrated AI-PLC frameworks; and iii) inadequate attention to the cultural and geographical specificities of frontier regions like Xinjiang.

This study seeks to address these gaps. Its contribution is threefold: i) Conceptual: It frames AI tools not as standalone solutions but as mediating artifacts within a sociocultural PLC framework. ii) Contextual: It designs and trials the model within the distinct urban-rural and multi-ethnic setting of Xinjiang. iii) Methodological: It employs a longitudinal, mixed-methods design to capture both quantitative outcomes and the richness of qualitative processes.

This study is anchored in Vygotskian sociocultural theory, which posits that learning is fundamentally a socially mediated activity. A central tenet is mediation, the process whereby cultural artifacts (including tools, signs, and language) intervene in human interactions with the world and with each other [6]. In this research, the AI-PLC platform and its constituent tools are conceptualized as digital cultural artifacts. They are designed to mediate interactions within the teacher community, facilitating collaborative problem-solving, peer scaffolding, and the externalization of teaching practice for collective analysis. This framework enables an examination of how technology fosters collaborative learning, moving beyond a simplistic assessment of whether it is effective.

Participants: A purposive sample of 42 secondary school English instructors (21 urban, 21 rural) from 12 intentionally paired “sister schools” in northern Xinjiang.Quantitative Data Sources:Pre- and post-intervention questionnaires assessing Teacher Self-Efficacy (adapted from an established scale) and Community Engagement (N = 42).Back-end platform analytics (login frequency, tool usage metrics, interaction logs).Qualitative Data Sources:Two rounds of semi-structured interviews (n = 20, approximately 30 minutes each).Monthly reflective journal entries submitted by participants (n = 150 entries total). Avoid combining SI and CGS units, such as current in amperes and magnetic field in oersteds. This often leads to confusion because equations do not balance dimensionally. If you must use mixed units, clearly state the units for each quantity that you use in an equation.

A six-month quasi-experimental intervention utilizing a convergent parallel mixed-methods design was conducted.

1. Baseline Phase (Month 1): Administration of pre-test questionnaires; orientation and training on the platform.

2. Intervention Phase (Months 2-6): Instructors engaged in structured weekly activities on the custom AI-PLC platform:

Activity 1: Co-annotation of lesson video recordings using an asynchronous collaborative tool.Activity 2: Discussion of AI-recommended pedagogical resources within themed online forums.Activity 3: Review and reflection based on automated, personalized classroom discourse analysis reports.Activity 4: Participation in peer dialogue sequences facilitated by a pedagogical chatbot.

3. Data Collection Phase (Ongoing): Continuous gathering of platform analytics; conduct of mid-point and final interviews; monthly collection of reflective journals.

4. Post-Intervention Phase (Month 6): Administration of post-test questionnaires.

Quantitative Analysis: SPSS 27.0 was utilized for descriptive statistics, reliability analysis (Cronbach’s α), and paired-sample t-tests to compare pre- and post-intervention scores.Qualitative Analysis: Interview and journal transcripts were analyzed using NVivo 14. A thematic analysis was performed, involving iterative cycles of open coding, axial coding, and theme development consistent with the mediation framework.

The design ensures methodological rigor through triangulation: methodological (multiple data sources), data (quantitative and qualitative), and theoretical (using a coherent framework for interpretation). The mixed-methods approach captures both the outcomes (quantifiable change) and the processes (lived experiences) of the intervention.

Feasibility was secured by: i) Partnering with existing school pairs to leverage pre-established relationships. ii) Implementing a custom platform built on stable, open-source components. iii) Obtaining informed consent and providing continuous technical support. iv) Designing platform activities with a manageable expected time commitment (approximately 1 - 2 hours per week).

The analysis of quantitative data revealed significant changes following the six-month intervention. The paired-sample t-test results for teacher self-efficacy are presented in Table 1.

Table 1.Paired-sample t-test of teacher self-efficacy (pre-test vs. post-test).

The data indicate a statistically significant increase in overall self-efficacy (p < 0.001). The most substantial gains were observed in the “Instructional Strategies” and “Student Engagement” subscales. While the mean increase for rural instructors (+0.62) was marginally higher than for their urban counterparts (+0.52), an independent samples t-test confirmed this difference was not statistically significant (p = 0.412).

Platform engagement metrics, illustrated in Table 1, demonstrated consistent weekly participation. The asynchronous annotation tool and the resource recommender were the most frequently utilized features.

Thematic analysis of interview transcripts and reflective journals yielded three primary themes:

1. Mediation of Collaborative Exchange: Digital tools, particularly the video annotation feature, established a shared reference point for dialogue. One urban instructor noted, “Commenting on the same video segment made our feedback concrete, moving beyond abstract advice.”

2. AI as a Catalyst for Introspection: Automated analytics, such as the classroom discourse report, served as powerful prompts for self-reflection. A rural instructor wrote, “The visualization showing my student talk ratio was a pivotal moment for reassessing my practice.”

3. Negotiation between Technology and Context: A prominent sub-theme involved adapting AI-suggested resources to fit specific classroom realities, such as multi-level or multi-ethnic cohorts, leading to collaborative “localization” strategies among participants.

The quantitative rise in self-efficacy, particularly in instructional strategies, is corroborated by qualitative accounts of acquiring new pedagogical ideas through mediated peer exchange. The high engagement metrics align with reports of diminished professional isolation. The challenge of “context negotiation” helps explain the comparatively smaller increase in “Classroom Management” efficacy, an area where the AI tools provided less direct, situation-specific support.

he results robustly support the hypothesis that an AI-PLC model can enhance instructor self-efficacy and collaborative practice. This aligns with established PLC literature [1] while extending it by demonstrating how AI tools can operationalize collaboration across distances. The finding that AI tools act as “catalysts for introspection” corroborates research on AI for feedback [3], applying it to a collaborative, instructor-focused context. The sustained engagement observed contrasts with some studies on online PLCs [2], suggesting that structured, tool-mediated tasks may mitigate common engagement declines.

The non-significant difference in efficacy gains between urban and rural instructors can be interpreted through the lens of mediation. The AI-PLC fostered a leveled communicative space where expertise was contingent on contribution rather than location, facilitating reciprocal rather than unidirectional learning.

1. For Educational Authorities: Investment should be directed towards AI-facilitated community platforms as core infrastructure for regional TPD. Policy should encourage sustained urban-rural school partnerships within such digital frameworks.

2. For Teacher Education Institutions: “AI-Mediated Collaboration” should be integrated into curricula to develop digital literacy for professional collaboration and evidence-based reflection.

Limitations include: i) constrained generalizability due to sample size and location; ii) inability to assess long-term sustainability; iii) measurement of perceived, rather than observed, changes in practice. Future research should: i) employ longitudinal designs to track impact on student outcomes; ii) develop “culturally responsive AI” algorithms [7]; iii) explore the model’s applicability to other disciplines and regions.