Abstract

The burgeoning field of online education has presented unique opportunities and challenges, particularly in resource-constrained regions like Africa. This paper delves into an innovative approach to bolstering online statistical training programs, addressing the quintessential issue of limited funding while ensuring high-quality training, with an application to an online training course on the System of Environmental-Economic Accounting. The crux of our research revolves around integrating cutting-edge large language models (LLMs), chatbots, and interactive knowledge graphs to simulate the dynamic and responsive experience of traditional tutor-led instruction. Our study is anchored in the context of Africa, where the demand for statistical education is on the rise. Still, the availability of resources and access to expert guidance remains constrained.
Our proposed methodology involves leveraging the capabilities of LLMs and language models to create an interactive, AI-driven educational environment. These models are designed to comprehend and respond to student queries, facilitate problem-solving, and provide personalized feedback, thereby emulating the nuanced interaction of a human tutor. Additionally, using interactive knowledge graphs in our system allows for a structured and visually engaging presentation of statistical concepts, enhancing the learning experience.
In this paper, we present proof of concept and discuss the efficacy of AI tutors in bridging the gap caused by the absence of human instructors in online, self-paced learning environments.
Furthermore, we explore the model's scalability, highlighting its potential to reduce the cost of online statistical training delivered by organizations in charge of statistical development in the continent with a minimal impact on quality.
Strand: Foundational work in statistics and data science