The exponential growth of artificial intelligence (AI) has revolutionized numerous fields, offering unprecedented capabilities in data analysis, prediction, and automation. These large language models (LLMs) can process, understand, and generate human language based on enormous amounts of data that are fed into it. AI is dependent on the types and nature of data that they are trained to access and process. LLMs are trained on large amounts of text, predominantly obtained from the internet to respond to questions, provide summaries, translate and create stories and other forms of textual outputs [1]. Nonetheless, AI forms much of its foundational operating system based on the existing reserve of knowledge on the internet. But is the existing resource of knowledge in the internet adequate, saturated and not to question, (fairly) representative of the population that are either underserved by the internet or are systematically deprived from the modern databases? This question already poses an input bias to these LLIMs and are indeed likely to generate bias in processing and output.