Teacher Professional Development on WhatsApp and LLMs: Early Lessons from Cameroon

Abstract: AI in education is commonly delivered through web-based systems such as online forms and institutional platforms. However, these approaches can exclude teachers in low-resource contexts, where everyday mobile platforms like WhatsApp serve as primary digital infrastructure. To address this gap, we present a field pilot in Cameroon that deploys a WhatsApp-based chatbot with LLM-supported content for teacher professional development (TPD), compared with an online form baseline. The system was evaluated through a mixed-methods study with 47 primary school teachers, integrating quantitative measures with qualitative insights from interviews and participant feedback. Results show that the chatbot was rated higher in perceived usability and overall experience, while learnability remained comparable. These improvements were driven by platform familiarity, low interaction overhead, and the modular structure of LLM-supported content, but were constrained by connectivity limitations, prepaid data costs, and multilingual needs (English/French). Building on these findings, we outline design directions for multilingual, culturally grounded interaction and for supporting prompting and reflection in AI use. More broadly, this work points to Thoughtful AI that supports reflection, relevance, and sustained professional growth.

• The paper demonstrates statistically significant usability gains with a WhatsApp-based, LLM-supported teacher PD intervention compared to traditional online forms.
• The mixed-methods approach, including surveys, chatbot logs, interviews, and reflections, rigorously evaluated user experience and engagement.
• The study identifies challenges such as connectivity issues, multilingual needs, and cultural misalignments, highlighting the need for iterative, localized AI design.

Teacher Professional Development on WhatsApp and LLMs: Early Lessons from Cameroon

Introduction

This study rigorously investigates the deployment of a WhatsApp-based LLM-supported chatbot for teacher professional development (TPD) in Cameroon, a context characterized by mobile-first access, intermittent connectivity, and bilingual (English/French) education. Drawing on a mixed-methods field pilot with 47 primary school teachers, the research directly compares chatbot-based TPD with traditional online forms, highlighting both the affordances and pitfalls associated with integrating AI-driven educational technologies into existing socio-technical infrastructures.

Context and Methodology

Cameroon's teacher workforce operates in resource-constrained settings where WhatsApp serves as the de facto platform for both professional collaboration and informal learning, embedded within daily communicative routines (Figure 1).

Figure 1: In-person teacher professional development workshop conducted as part of the study, contextualizing the WhatsApp-based TPD intervention.

The pilot intervention, conducted during a week-long, bilingual workshop, introduced teachers to a WhatsApp-based chatbot delivering modular, pedagogy-aligned content. Notably, lesson authoring and content variation incorporated LLM support through a facilitator dashboard that streamlined content creation and review workflows (Figure 2).

Figure 2: Instructor dashboard for authoring and managing WhatsApp-delivered TPD with analytics and real-time preview capabilities.

Data collection comprised pre- and post-intervention surveys, chatbot logs, school-based group reflections, and semi-structured interviews, triangulating quantitative measures with qualitative accounts to capture user experiences, usability, learnability, and perceptions of AI in TPD.

System Description

The WhatsApp chatbot delivered professional learning via structured, conversational modules, integrating reflection prompts, micro-content, and multimedia elements directly within users' primary communication channel (Figure 3).

Figure 3: Chatbot lesson flow illustrating course navigation, embedded reflection, and module progression within WhatsApp.

LLMs facilitated content variation and prompt generation, but final output was moderated and adapted for local context by NGO facilitators. Content modules were directly aligned with the TPD provider's established frameworks, encompassing evidence-based strategies such as relationship-centered classroom management, play-based learning, and emotion regulation heuristics.

Comparative Evaluation: WhatsApp Chatbot vs. Online Form

The WhatsApp-based intervention demonstrated statistically significant gains in usability and overall experience over the online form baseline (usability: $M_{chatbot} = 7.82$ vs. $M_{form} = 7.33$, $p = .046$; combined scale: $M_{chatbot} = 7.41$ vs. $M_{form} = 7.03$, $p = .044$). Both modalities exhibited comparable learnability.

Key determinants of enhanced chatbot usability included:

• Platform familiarity: Over 87% of participants owned smartphones; WhatsApp was already central to daily routines, minimizing onboarding friction.
• Low interaction overhead: Stepwise, menu-driven engagement aligned with habitual practices, in contrast to forms requiring additional scaffolding.
• Scripted modularity: Content pacing and embedded reflection supported iterative use and situated transfer to classroom practice.

These design features resulted in increased teacher autonomy and willingness to return to the resource, with several teachers describing the chatbot as a persistent reference for pedagogical strategies.

Challenges and Contextual Barriers

Despite usability improvements, the intervention surfaced persistent challenges:

• Connectivity and cost: Data scarcity, unreliable networks, and prepaid models repeatedly disrupted access and module completion, overshadowing interface-level affordances.
• Multilingual needs: English-only delivery excluded Francophone teachers, emphasizing the necessity of robust multilingual and localized content pipelines.
• Rigid conversational flows: Teachers requested more flexible, open-ended interactions—current menu-driven logic limited personalization, spontaneity, and nuanced pedagogical dialogue.
• Cultural specificity: Some LLM-generated content was perceived as culturally misaligned (e.g., references to Western inclusion paradigms rather than local disability inclusion norms), underscoring the need for participatory localization and community-driven data curation.

Implications for AI and TPD in Low-Resource Settings

The empirics from this deployment underscore several implications for the design and deployment of AI-powered TPD:

• Thoughtful AI Integration: Effective LLM-supported TPD requires scaffolding for both teacher prompt engineering and critical reflection, not just content delivery. Teachers should be empowered to iteratively refine prompts and encouraged to use AI outputs as a basis for reflective practice.
• Localized, Multilingual Models: Addressing multilingual and cultural adaptation gaps is imperative. Community-driven dataset development and participatory design should be core to ongoing system development.
• Support for Long-Term Growth: Teachers articulated aspirations beyond immediate task support, seeking AI support for career planning, goal-setting, and exposure to broader educational paradigms. Adaptive features, such as scenario-based simulations and goal tracking, are recommended to support professional growth trajectories.
• Scalability and Evaluation: To ascertain sustained impact, transitions from pilot deployments to longitudinal, at-scale studies are essential. Future research should focus on real-world engagement, iterative content refinement, and rigorous outcome measurement over extended timelines.

Conclusion

The findings from this study provide robust evidence that WhatsApp-based, LLM-supported TPD modules yield higher usability and improved experience relative to web-based forms in mobile-first, resource-constrained settings. However, infrastructural, linguistic, and localization challenges persist and necessitate ongoing investment in platform adaptation, data pipeline pluralism, and thoughtful pedagogical integration. Continued iteration focused on reflection-driven AI use, inclusivity, and support for teacher aspirations will be required to translate such pilots into sustainable systems for TPD at scale.