Evolving Education in India: Multilingual LLMs for Vernacular Learning Experiences

The recent advancements in Artificial Intelligence (AI), particularly in the domain of natural language processing (NLP), have brought in an era of technological innovation. At the forefront of this are Large Language Models (LLMs). LLMs are neural network-based transformers trained on vast datasets. Due to their massive training data, they imbibe the intricacies of human conversation well and understand and generate human-like text. LLMs have the potential to revolutionize various sectors, with applications ranging from machine translation to sentiment analysis amongst other advanced NLP tasks. In the context of the diverse educational landscape of India, technical integration has become of the utmost importance in enhancing learning experiences and accessibility to relevant material. Amongst the many tools available to aid education, LLMs emerge as transfor-mative assets, reshaping classroom dynamics as well as bridging linguistic divides. This research explores the comprehension abilities of GPT 3.5 Turbo in Hindi, a language widely spoken across the Indian subcontinent. After use-case specific training with reading comprehension passages, a set of multiple-choice questions (MCQs) are used to assess the performance of the large language model. With a focus on vernacular languages, this study aims to impact a significant portion of the Indian student population, contributing to the availability of specialized education technology in a multilingual context.