Teaching and Working with EthioLLM: A Proposal to Build Ethiopia’s AI Pipeline from Classroom to Workforce

Abstract

In 2020, the Federal Democratic Republic of Ethiopia published “Digital Ethiopia 2025,” which is a 150-page agenda set forth to transform the nation through the use and expansion of various digital technologies. I refer to it hereafter as the Agenda. As 2025 nears its end, the Agenda has fallen short of several key objectives, including improving primary education and expanding the digital workforce. In this paper, I examine the Agenda’s discourse and implementation in order to shed light on the failures of foreign frameworks for local issues and the consequences for the Ethiopian working class. Ultimately, I argue that by centering African epistemologies and implementing culturally grounded and project-based technical primary education, Ethiopia can establish an internationally competitive AI workforce. My first recommendation is a primary school curriculum introducing foundational AI concepts through EthioLLM, a large language model that uses indigenous Ethiopian languages and is not yet widely deployed. This curriculum supports empowerment through mother-tongue instruction and cultivates ownership over local problem-solving. My second recommendation is the creation of a digital platform cooperative that specializes in EthioLLM-based services in order to provide a competitive alternative to traditional Business Processing Outsourcing (BPO) models which often exploit the Global South. This repositions the Ethiopian working class as equal partners and co-creators of value rather than extractable labor. Together, these interventions offer a lasting model for an AI pipeline that honors African epistemologies and local priorities every step of the way, from the classroom and workforce to the global digital economy.