The new Hindi large language model

Just when you thought AI news and developments have reached peak saturation, it seems there’s always something new that captures our collective imagination. Every week brings a new large language model – the backbone of many AI implementations. This time, AI is getting a Hindi-language breakthrough. It’s part of NANDA, a project unveiled earlier this month at the UAE-India Business Forum in Mumbai. Named after one of India’s highest peaks, NANDA is a technological – and cultural – feat in itself that’s supposed to open the doors to AI for more than half a billion Hindi-language speakers. NANDA is a partnership between G42, the UAE-based tech holding group, the Mohamed bin Zayed University of Artificial Intelligence and the American AI company Cerebras. G42 describes the project as a cutting-edge Hindi Large Language Model – trained on one of the most advanced supercomputers in the world. That’s a departure from the usual news about AI LLMs, which for a long time, have been centred around the English language, leading some to worry about an uphill battle for other languages and cultures trying to get a foot into artificial intelligence. So, this is a monumental step forward in artificial intelligence – and broader tech accessibility. But in a world where mentions of AI conjure concerns of tech ethics and information wars, where do up-and-coming LLMs fit in? In this episode of Business Extra, host Cody Combs hears from Preslav Nakov, a professor at MBZUAI specialising in artificial intelligence and computational linguistics, about what this means for linguistic and cultural diversity – or whether we're just scratching the surface of what’s possible.