Mitesh Khapra, Championing Indian Languages in AI, Makes TIME100 AI List

Mitesh Khapra realised quite some time ago that technical development in Indian languages is lagging behind because of a lack of sufficient data for these languages.

Indians and their intellect are renowned worldwide, a fact repeatedly proven. Another such instance is making headlines. Time Magazine recently released its "TIME100 AI" list for 2025, featuring 100 leading tech CEOs, founders, and innovators globally. This prestigious list includes prominent figures like Tesla CEO Elon Musk, OpenAI CEO Sam Altman, and Meta CEO Mark Zuckerberg, alongside Indian professor Mitesh Khapra, who has played a crucial role in making Artificial Intelligence (AI) accessible to ordinary people in Indian languages.

Mitesh Khapra is an Associate Professor in the Department of Computer Science and Engineering at IIT Madras. He began his career as a software engineer at companies like Infosys and LG. Subsequently, he held research positions at Microsoft India and IBM before joining IIT Madras in 2016.

Mitesh Khapra also founded One Fourth Labs, aiming to provide high-quality, affordable AI education in India to strengthen AI for society and industry. His expertise lies in deep learning, natural language processing (NLP), and conversational systems.

Mitesh Khapra: Focus on AI for Indian Languages

According to Time, Khapra recognised early on that technical progress in Indian languages lagged due to insufficient data. While Western AI models perform well in Hindi and Bengali, they falter with less prevalent Indian languages. To address this, under his leadership, a research lab called AI4Bharat recorded voices from diverse backgrounds across nearly 500 districts in India, creating a robust dataset for the country's 22 official languages.

AI4Bharat's Excellent Work

Established in 2019, AI4Bharat is now an official partner in the Indian government's "Bhashini Programme," aimed at providing citizens with digital services in their native languages. AI4Bharat supplies 80% of the data for this project, and this data is open-source, allowing other developers to use it for training their AI models. Mitesh Khapra states, “If a large tech company uses this data to improve its models in Hindi or Marathi, the entire country will benefit.” AI4Bharat's AI models are being used for document translation in the Supreme Court, and a voice bot has been created for farmers to report issues related to government subsidies.