<\/p>\n
NEW DELHI<\/strong> : Bhashini, India\u2019s national language database platform, has begun showcasing beta-stage deployment of its multi-model generative artificial intelligence services, a step towards taking the benefits of the technology to the vast majority of non-English-speaking populations.\u00a0<\/p>\n<\/p><\/div>\n Demonstrating its abilities at the 2023 Mint AI Summit on Friday, Amitabh Nag, chief executive of Bhashini\u2014an undertaking of the ministry of electronics and information technology (Meity)\u2019s National Language Translation Mission\u2014demonstrated the ability of Bhashini in conducting voice-driven contactless digital payments through UPI.<\/p>\n It was a demonstration of how generative AI, or GenAI, can contribute as a digital public good, and how the technology infrastructure would be crucial to the overall deployment of services in empowering various communities across economic strata.<\/p>\n Speaking about the deployment of GenAI in the Centre-backed Bhashini platform, Nag said that a key part of these services would lie in building purpose-driven AI models.\u00a0<\/p>\n \u201cThe specific use cases of generative AI are very narrow, and so, their datasets have to be narrow as well for us to deploy them as services very fast. The more use cases we have, the better the situation would be in terms of deriving outcomes and facilitating adoptions,” Nag said.<\/p>\n A key challenge to this process, however, is the availability of digital datasets in regional languages.\u00a0<\/p>\n Addressing this issue, Nag said, \u201cMany of the Indian languages don\u2019t have digital data. As a result, we spend around $6-7 million to collect the data.”<\/p>\n This step is also crucial in personalising services, the starting point of which begins with language databases, Nag added.\u00a0<\/p>\n The Bhashini mission, in a bid to build this up, has also introduced a crowdsourcing model to collect data\u2014called Bhasha Daan.<\/p>\n Bhashini is a part of the India AI mission\u2019s datasets platform, which seeks to build organised databases to collect and make local language data available for enterprises and researchers alike.\u00a0<\/p>\n Nag elucidated that multiple AI models are being rendered as part of the Bhashini platform, which can do \u201cabout 100,000 inferences per day, and can be scaled up to 500,000 inferences.”<\/p>\n All of this will be crucial in bridging India\u2019s digital divide, Nag said. Adding to this, he added that speech interfaces and multi-modal input interfaces would help bridge both digital and literacy divides in the long run.<\/p>\n Milestone Alert!<\/strong>Livemint tops charts as the fastest growing news website in the world \ud83c\udf0f<\/span> Click here<\/strong><\/a> to know more.<\/p>\n