From palm manuscripts to printing press and later, websites… how we store and read information has changed drastically in the span of a few centuries. The latest in that natural evolution is large language models (LLMs).
Heeding this transition and perhaps hoping to capitalise on the state’s youngsters’ proclivity towards everything technology, the Kerala government has invited developers and innovators to build and train a high-quality language model in Malayalam.
In the latest budgetary allocations, Rs 1 crore has been earmarked for this initiative.
“While global AI tools can support Malayalam to an extent, the quality and consistency required for Malayalam-first uses (learning, communication, content creation and digital services) remain uneven,” said Jyothis K S, co-founder of Zappyhire in Kochi.
This is because systems are largely designed and optimised for larger, English-first contexts, he said, adding, “They don’t understand the nuances and the sub-text of Malayalam.”
A Malayalam-first model, Jyothis highlighted, does three things. “One, it safeguards our language, and by extension, our culture. Two, it empowers the modern Malayali to translate their ideas using their mother tongue and not lose substance due to system’s shortcomings. Three, this digital leap, no doubt, will usher in a revolution in the language by expanding its acceptability and accessibility,” he explained.
The move builds on the state’s ongoing efforts to strengthen AI infrastructure and innovation, whilst also encouraging serious research and product development in Malayalam, the budget note read. “This LLM push is indeed a welcome move, and an essential one,” Jyothis concluded.