iit: IIT Madras to pilot courses in nearly a dozen languages as part of Bhashini push

Bhashini aims to build a national public digital platform for languages to develop services and products for citizens, using artificial intelligence and other emerging technologies.

IIT-M director V Kamakoti told ET that while India is going through a massive wave of digitalisation, language can be a hurdle. He said that if something is disseminated in only a set of languages, people who do not know those languages cannot access the content.

“Digital access may be provided but there is an extreme necessity to see that there is a seamless language transition,” Kamakoti told ET. “Bhashini is a big LLM (large language model) which will be helpful in this. We are looking to pilot it within the campus by this financial year by providing students with access to around 10-15 courses in about 12 of the 22 official languages. Once this is done, we’ll go to add more courses and languages.”

The goal is to have two people who do not know a common language being able to converse through an intermediate LLM-driven technology which will, on the fly, translate into the languages they know.

Also read | Talk to ‘Bhashini’ for UPI-based services in local languages

Discover the stories of your interest

Kamakoti said it was essential for all important public digital initiatives and digital infrastructure to be available in all official languages of the country.”Today, we would like to have a ‘BharatGPT’ Large Language Model, which also essentially depends upon language translation,” Kamakoti said. “We need BharatGPT-like services for working with public services like Department of Administrative Reforms and Public Grievances, which is a public digital infrastructure given to the people for recording their grievances. They must be able to record in any language of their choice. The person addressing the grievance can take it in any language of her/his choice and give back a response and act,” he said.

The director also spoke about the institute’s work to help access to justice by making it more linguistically inclusive, while also shedding light on the work that the IIT was doing to empower lawyers and judges to be assisted effectively in not just their preparation for a case but also while delivering judgements and verdicts.

Also read | Bhashini to launch RBI’s platform for frictionless credit in multiple languages

Petitioners or respondents need not be conversant in English but should be comfortable in presenting their case in their mother tongue or a language of their choice, he said, adding that the BharatGPT-type of translation model would help a lot with that.

“BharatGPT-type of software can be used to summarise information, like helping a lawyer summarise multiple judgements and can become a ‘co-pilot’ to a lawyer, providing assistance in writing briefs. BharatGPT can also help the honourable judges to get summaries of large briefings,” Kamakoti said. “We can develop such types of co-pilots in multiple languages. These are just two case studies. There can be multiple such case studies where BharatGPT can find major use.”

IIT-M’s work in this regard follows the Supreme Court signing a memorandum of understanding with the institute on October 12 to collaborate on using AI and emerging technologies for transcription and translation tools, summarisation of page transcript, exclusive streaming platform for court trials, process automation and large languages models. The pact was signed after Chief Justice of India DY Chandrachud’s visit to IIT-Madras in July this year.

Bhashini aims to build a national public digital platform for languages to develop services and products for citizens, using artificial intelligence and other emerging technologies.

IIT-M director V Kamakoti told ET that while India is going through a massive wave of digitalisation, language can be a hurdle. He said that if something is disseminated in only a set of languages, people who do not know those languages cannot access the content.

“Digital access may be provided but there is an extreme necessity to see that there is a seamless language transition,” Kamakoti told ET. “Bhashini is a big LLM (large language model) which will be helpful in this. We are looking to pilot it within the campus by this financial year by providing students with access to around 10-15 courses in about 12 of the 22 official languages. Once this is done, we’ll go to add more courses and languages.”

The goal is to have two people who do not know a common language being able to converse through an intermediate LLM-driven technology which will, on the fly, translate into the languages they know.

Also read | Talk to ‘Bhashini’ for UPI-based services in local languages

Discover the stories of your interest

Kamakoti said it was essential for all important public digital initiatives and digital infrastructure to be available in all official languages of the country.”Today, we would like to have a ‘BharatGPT’ Large Language Model, which also essentially depends upon language translation,” Kamakoti said. “We need BharatGPT-like services for working with public services like Department of Administrative Reforms and Public Grievances, which is a public digital infrastructure given to the people for recording their grievances. They must be able to record in any language of their choice. The person addressing the grievance can take it in any language of her/his choice and give back a response and act,” he said.

The director also spoke about the institute’s work to help access to justice by making it more linguistically inclusive, while also shedding light on the work that the IIT was doing to empower lawyers and judges to be assisted effectively in not just their preparation for a case but also while delivering judgements and verdicts.

Also read | Bhashini to launch RBI’s platform for frictionless credit in multiple languages

Petitioners or respondents need not be conversant in English but should be comfortable in presenting their case in their mother tongue or a language of their choice, he said, adding that the BharatGPT-type of translation model would help a lot with that.

“BharatGPT-type of software can be used to summarise information, like helping a lawyer summarise multiple judgements and can become a ‘co-pilot’ to a lawyer, providing assistance in writing briefs. BharatGPT can also help the honourable judges to get summaries of large briefings,” Kamakoti said. “We can develop such types of co-pilots in multiple languages. These are just two case studies. There can be multiple such case studies where BharatGPT can find major use.”

IIT-M’s work in this regard follows the Supreme Court signing a memorandum of understanding with the institute on October 12 to collaborate on using AI and emerging technologies for transcription and translation tools, summarisation of page transcript, exclusive streaming platform for court trials, process automation and large languages models. The pact was signed after Chief Justice of India DY Chandrachud’s visit to IIT-Madras in July this year.