Between Promise and Reality: Navigating the Landscape of Indic Language LLMs Amidst Hyperbole and Hope

# 186






Abstract

The emergence of Large Language Models (LLMs) tailored for Indic languages marks a significant milestone in the quest for linguistic inclusivity in the digital age. However, this journey is fraught with both groundbreaking achievements and contentious claims. This talk aims to dissect the current landscape of Indic language LLMs, critically examining the dichotomy between the ambitious promises made by various companies and the tangible progress achieved. We initiate our exploration by addressing the complex challenges in developing LLMs for the diverse and nuanced spectrum of Indic languages, including issues related to script variability, linguistic heterogeneity, and the digital scarcity of many languages. By showcasing case studies and empirical evidence, we seek to demystify the hype, presenting a balanced view of where Indic language LLMs currently stand, their real-world applications, and their limitations. Furthermore, we engage with the broader implications of these developments, pondering their impact on society, culture, and the digital economy in the Indian subcontinent. As we chart a course towards the future, our discussion emphasizes the need for transparency, ethical AI development, and robust community engagement.

Dr. Gaurav Aggarwal, Founder and CEO of Ananas Labs

A Computer Science BTech from IIT Madras and PhD from University of Maryland, Gaurav has over been working in the area of Machine learning for nearly two decades including successful stints at Google Research, Yahoo Labs and his own startup Fashiate which was acquired by Snapdeal. Gaurav is currently the Founder and CEO of Ananas Labs bringing fundamental advances in AI, through real-world applications. He is also an anchor volunteer at iSPIRT, a mission-oriented, nonprofit, technology think-and-do tank, driving 30 years orbit shifts. At iSPIRT, Gaurav is actively involved in bringing techno-regulatory innovation for AI in India.