Developing an AI chatbot to facilitate teacher training in SEL kernels

• In Event: Leveraging Technology for Social and Emotional Learning at Scale

Proposal

In recent years, programs targeting children’s development in social and emotional learning (SEL) have gained considerable attention in international education (Corcoran et al., 2018). In resource-abundant settings, SEL programs have been found to enhance student well-being, improve academic outcomes, foster positive peer relationships, and create a supportive school climate (Cipriano et al., 2023).

In low- and middle-income countries (LMICs), the last decade has also seen a marked increase in SEL programs (Deitz et al., 2021). Among these initiatives, SEL kernels—flexible, frequent, and applicable activities that are easily integrated into regular classroom practices—have shown particular promise (Jones et al., 2017). Preliminary evidence suggests that they can have a positive and promising impact on children’s social-emotional and academic development in low-resource settings (Brown et al., 2023; Colagrossi et al., 2024). However, despite the growing presence of SEL programs in LMICs, substantial challenges remain that hinder their widespread adoption and effectiveness.

One of the primary challenges is the absence of a universal framework for cultural adaptation. Organizations mostly work in isolation to develop their own SEL programs and lack cross-contextual collaboration or adaptation (Dhata & Swati, 2020). Furthermore, the time and financial costs of implementing SEL programs at scale, particularly those requiring in-person training, are prohibitive in many LMICs. Teachers are often overburdened with numerous responsibilities and find it difficult to engage in additional training or to implement complex new programs (Evans & Yuan, 2018). Finally, there is a significant gap in the evidence base regarding which SEL programs are effective in low-resource settings, for whom they work, under what conditions, and through what mechanisms (Aber et al., 2021).

In response to these challenges, we at the Agency Fund propose the use of generative AI (GenAI) to facilitate the training of teachers in culturally relevant SEL kernels. Specifically, we aim to develop and evaluate a GenAI-powered chatbot to address the barriers to SEL implementation in low-resource settings. This chatbot would offer a novel approach to teacher training that is scalable, adaptable, and capable of meeting the unique needs of teachers in diverse cultural contexts. First and foremost, large language models (e.g., GPT-4o) have powerful real-time translation and cultural adaptation capabilities, even with minimal input. With access to an open-sourced chatbot powered by these models, organizations can readily create and evaluate culturally relevant SEL kernels within a shared theoretical framework. Second, a GenAI chatbot can provide a flexible and engaging learning experience for teachers who are already multitasking. Rather than requiring time-consuming, in-person training sessions, the chatbot can deliver interactive, immersive learning experiences through multi-media conversational exchanges that are likely easier for teachers to integrate into their daily routines. Furthermore, the chatbot can be deployed through widely used platforms such as WhatsApp or mobile apps, allowing it to reach a broad audience of educators with minimal costs. Chatbots can also naturally incorporate automatic data collection, text analysis, and survey tools to gather real-time feedback and insights from teachers about their learning and teaching experiences.

Author

• Zezhen Wu, The Agency Fund