Implementation learnings from AudioClase in Colombia

• In Event: What works to support continuous learning when schools close through personalized learning?

Proposal

The COVID-19 pandemic has exacerbated existing problems in the educational system, and Colombia is no exception. In Colombia, all in-person education programs were suspended in April 2020, and they began to reopen gradually until 2021. The Colombian Ministry of Education quickly pivoted and digitized educational content to mitigate learning loss to provide educational opportunities at home, mainly via the use of computers, and delivered by internet. However, there were few programs for youth between the ages of 12 and 17 in comparison with other age groups.
To respond to this gap in the education provision, as well as challenges caused by the pandemic (such as inadequate content for self-learning, limited teacher-student relationships, and limited access to stable internet, as highlighted by research and a 2-week virtual design sprint conducted by IRC and partners in Colombia (Parolin et al, 2021; IRC, 2022)), the IRC developed Audio-Class System (ACS), an Interactive Voice Response (IVR) for feature phones and WhatsApp Chatbots for smartphones. Using AI technology, the ACS platform allows for automated supportive messaging, automated lessons, tailored content, and an alternative to enable two-way communication between teachers and students. ACS uses audio as its main educational media because audio can be shared across all devices, be it radio, feature phone, smartphone, or tablet. to deliver a range of educational content.

In this session will present findings of ACS implementation research in Bogotá and Medellín, Colombia, the first full round of implementation after a small pilot. The program implementation consists of tailored curriculum, a curriculum-based assessment, chatbot flows and assets - audios, visuals and texts, dashboard data, printed materials and a training/technical assistance structure for teachers and community tutors. Using a sequential mixed methods design (Cresswell, 2014), this research study examines the implementation of the ACS pilot Year 2. In this presentation we will provide insights from qualitative and quantitative data, to examine the quality and fidelity of implementation, uptake and use, users' perceptions of the functionality appropriateness, and emerging findings on student learning. The session will also highlight how learnings and lessons learned when implementing a personalized AI-based flexible low-tech solution. Conclusions and lessons learned will help build the still relatively scant evidence on these Ed-Tech solutions, as well as others experiencing similar challenges.

While applications of AI in education to education are emerging at a rapid pace, studies on the effectiveness of specific programs and initiatives are scarce, with most evidence emerging from the medical and higher education fields (Engzell, 2020; Villegas et al., 2020). A recent systematic review of AI-based robots in education (Chu, et al, 2022) found that most research focuses on learners of elementary and higher education levels, analyzed experiments within or less than 4 weeks in a physical environment, and in the disciplines of Language and Science. Our research will thus contribute to the literature by researching student learning of multiple subjects, using a sample of secondary learners and teachers over a longer exposure period (27 weeks), and exploring changes in use, self-efficacy, and satisfaction

Authors

• Idalia Rodriguez, International Rescue Committee

• Camilo Valenzuela, International Rescue Committee IRC