Education Minister Dharmendra Pradhan last week said the Centre will integrate AI tools in teaching from the next academic session at all levels, from pre-primary to higher education.
The effort is propelled by the launch of a not-for-profit company, Bodhan AI, at the Centre of Excellence in AI for Education last week. IIT Madras hosts this Centre of Excellence, which was a Budget announcement last year, with an allocation of Rs 500 crore.
Here’s how the Centre plans to involve AI tools in education through Bodhan AI.
What is Bodhan AI going to do?
The company will develop the Bharat EduAI Stack as Digital Public Infrastructure (DPI) for education. DPI is a digital system that can provide access to services on a large scale, like UPI (Unified Payments Interface) for payments.
The company will work on research to build AI capabilities for Indian languages, and develop assets like automatic speech recognition and speech synthesis. Applications will then be built, with the aim of taking them to schools and other institutions through collaborations with state governments.
With this, it hopes to work with edtech companies.
“Putting these assets in the open by itself is not going to solve the problem…These models have to be deployed on a sovereign infrastructure where now you can just plug in these models into your existing edtech solutions. You already have an existing edtech solution for measuring oral reading fluency in English. If, at Bodhan AI, I have developed a speech diagnostics model for Hindi, can you just plug that into your existing solution and now start distributing it to Hindi medium schools…We can host it on an AI infrastructure and try to provide it at cost,” Prof Mitesh Khapra from IIT Madras said during a conclave held by the Education Ministry this week.
He added: “On top of this, we as well as others can build applications. This is where the tech players come in. You build solutions, integrate our models, our technology with your solutions, enhance them, and distribute. We’ll do randomised control trials in a large number of schools… The last mile delivery is where we would rely on government partners, the different education bodies, both within the Union government and state governments. There are many tech players today building these solutions. But they are all plugging in models that are not Indian. They are relying on the ChatGPTs or the Geminis of the world”, Khapra said.
What sort of AI tools are likely?
Prof V Kamakoti, Director, IIT Madras, said that the focus of the tools will be on personalised learning for the student, assistance for teachers and parents to make interventions, and for the administration to determine how districts or schools might be faring and take policy decisions.
“As a child, I need to understand…in my mother tongue…where I am going wrong, so that I can correct myself. We want AI to be able to do this,” Kamakoti said.
For the student, this can mean voice-based exercises on a phone, tablet, or laptop, with feedback or reports that will be provided to the teacher. The AI tool can then recommend exercises or personalised worksheets that the teacher can use with the student for improvement.
“If the child asks a question, the model should answer. This is all about language models. The way AI can help in education is predominantly language…the ability to understand and interpret that language. It should be multilingual,” Prof Kamakoti said.
Several such tools already exist. What next?
With the conclave last week, IIT Madras examined the existing landscape of edtech companies, their work and reach, Prof Kamakoti said.
“What the Bodhan AI company will now do is build the basic blocks. A sovereign model will be built –the digital public infrastructure. If one of these edtech companies wants to deploy a tool, we will have the basic infrastructure for them, and they can build on it. We are to become a basic technology provider, and application aggregator. Then we go to a state government or district, and these models can be deployed in schools. Impact can be studied,” he said.
In addition to the funds that have already been allocated in the budget, funds will come in through the start-ups and governments that are using the system. “The state governments will pay some maintenance costs. The start-ups will also give some equity in return for the services. After a point, it will become like the Linux operating system…taken over by the community,” Prof Kamakoti said.
What are the concerns?
“We should be very careful about the data. Anything that the student puts in or writes is their personal data. We need to see that there is absolutely no data storage in a public forum. But it can be stored within a device that the student will use,” Prof Kamakoti said.
As for screen use, he pointed to the use of voice-based tools.
While Khapra said that “we will be grounded in the government policy framework of the NEP 2020”, Karthik Mahadevan Mohanakrishnan, the COO of Bodhan AI, added that the research and tools will be based on NCERT and SCERT frameworks.
Non-profit Bodhan AI will develop the Bharat EduAI Stack as Digital Public Infrastructure for education. (Photo: bodhan.ai) 
Share
