Are LLMs Useful in the Poorest Schools? TheTeacher.AI in Sierra Leone
Project Overview
The document explores the use of generative AI in education, specifically through the implementation of TheTeacher.AI, an AI chatbot designed to support teachers in Sierra Leone with professional development and instructional assistance. It addresses the significant challenges faced by educators in low-resource environments, such as limited access to training and resources. The chatbot provides on-demand support via WhatsApp, enabling teachers to receive timely help regardless of their location. The implementation included comprehensive training sessions aimed at enhancing teachers' ability to utilize the tool effectively. Despite encountering obstacles like inconsistent internet connectivity and diverse levels of digital literacy among users, TheTeacher.AI has achieved sustained engagement and garnered positive feedback from educators. The findings indicate that generative AI can play a crucial role in enhancing educational practices in underserved regions, ultimately fostering professional growth and improving teaching outcomes.
Key Applications
TheTeacher.AI chatbot
Context: Assisting teachers in Sierra Leone with professional development and instructional support.
Implementation: Implemented through WhatsApp, with teachers sending queries that are processed by the AI.
Outcomes: Improved teacher engagement in lesson planning, classroom management, and subject matter understanding.
Challenges: Limited internet access, varying levels of digital literacy, and the need for contextual relevance in responses.
Implementation Barriers
Technical barrier
Poor and intermittent network connectivity in Sierra Leone hampers the use of the chatbot. Limited technical competence among teachers makes it challenging for them to use the chatbot effectively.
Proposed Solutions: Utilizing WhatsApp for its low data consumption and better performance under poor conditions. Providing targeted training sessions to help teachers learn how to ask effective questions.
Financial barrier
High costs associated with using GPT models relative to low incomes in Sierra Leone.
Proposed Solutions: Focusing the tool's use on teachers rather than students to prioritize educational contexts.
Project Team
Jun Ho Choi
Researcher
Oliver Garrod
Researcher
Paul Atherton
Researcher
Andrew Joyce-Gibbons
Researcher
Miriam Mason-Sesay
Researcher
Daniel Björkegren
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Jun Ho Choi, Oliver Garrod, Paul Atherton, Andrew Joyce-Gibbons, Miriam Mason-Sesay, Daniel Björkegren
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai