Scaling Down to Scale Up: Small Language Models for Large Scale Educational Impact

This conference talk explores how Small Language Models (SLMs) can revolutionize education in resource-constrained settings. Learn how these lightweight, open-source AI systems offer advantages over Large Language Models (LLMs) by requiring fewer computational resources while still delivering personalized learning experiences. The speakers, Anindita Sinha Banerjee and Abhijit Roy, demonstrate how SLMs can be fine-tuned on localized data, integrated into existing learning platforms, and deployed in low-resource environments to create adaptive educational tools. Discover the key strengths of SLMs including affordability, privacy protection through on-device processing, support for local languages, and energy efficiency. The presentation includes a live demonstration of a personalized learning assistant that adapts to individual student needs while remaining accessible on basic hardware. While acknowledging limitations in handling complex tasks, the talk provides a practical roadmap for implementing SLMs to democratize quality education across diverse settings.