Examine the Opportunities and Challenges of Large Language Model (LLM) For Indic Languages

Large Language Models like GPT and BERT have made significant advancements in NLP, particularly in text generation, translation, and summarization. However, their application in Indic languages remains relatively unexplored due to unique linguistic challenges such as complex morphology, diverse scripts, and limited digitized resources. This systematic literature review follows PRISMA guidelines to identify, analyze, and evaluate existing research on the opportunities and challenges of LLMs for Indic languages. The review covers relevant publications from databases like Web of Science, IEEE Xplore, and Springer. Inclusion criteria were applied to studies published between 2000 and 2024, with a focus on LLM architectures like GPT and BERT as applied to Indic languages. From 161 selected articles, this review highlights the potential of LLMs in improving machine translation, speech recognition, and cross-lingual NLP tasks for Indic languages. The review reveals that while LLMs show promise in enabling better language processing for Indic languages, challenges persist, particularly the scarcity of annotated datasets, script diversity, and the computational resources required for training. LLMs offer great potential to democratize access to digital content for Indic languages, helping bridge the linguistic digital divide. However, the challenges of limited data, linguistic diversity, and bias require collaborative research and tailored solutions. Comparison of Large Language Models for Indic language and in general also made a in this paper. This review identifies gaps in the current literature and suggests future directions to improve the efficacy of LLMs for underrepresented languages.