Union Minister Jitendra Singh Emphasises Ethics in Health AI as BharatGen Launches Param 2

Singh noted that BharatGen represents a government-owned sovereign model but acknowledged that private players are also active in digital health and AI-driven documentation.

Union Minister Jitendra Singh has underscored that ethical safeguards will determine how effectively technology such as BharatGen transforms India’s health ecosystem, as it launches Param 2 at the India AI Impact Summit 2026.

The 17-billion-parameter model, designed as a mixture-of-experts architecture, supports 22 Indian languages and has been positioned as a foundation for healthcare, governance and education applications.

BharatGen, developed under the National Mission on Interdisciplinary Cyber-Physical Systems and supported by the Department of Science and Technology, was first introduced in 2024 as India’s government-funded multilingual and multimodal AI initiative.

Backed by INR 235 Cr in early-stage funding and further strengthened under the India AI Mission, the programme has aimed to reduce reliance on foreign AI systems and create models trained on India-centric datasets.

The newly launched Param 2 builds on earlier versions and domain-specific models such as Ayur Param for Ayurveda. In healthcare, BharatGen’s models are expected to assist with clinical documentation, speech-to-text for medical consultations, multilingual patient communication, digital case histories and telemedicine workflows.

Its Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) systems are particularly relevant for public hospitals and primary health centres serving linguistically diverse populations.

Speaking at the summit, Jitendra Singh stated, “Ethics will be a very important determinant of how optimally AI can be used in the health sector.” He noted that BharatGen represents a government-owned sovereign model but acknowledged that private players are also active in digital health and AI-driven documentation.

“In the times to come, we would also have to integrate. Neither can they work in isolation, nor can we. So, we have to be prepared for that,” he said.

BharatGen CEO Rishi Bal highlighted that most real-world deployments do not require trillion-parameter systems. “A majority of business, government and many consumer cases actually don’t need a trillion-parameter frontier model,” he said, adding that domain-focused, locally adaptable models may deliver stronger outcomes in emerging markets.

Param 2 has been trained using thousands of GPUs under government-backed compute allocations. The model is being released as a pre-trained checkpoint, enabling institutions to fine-tune it for specific needs, including health records analysis, regional language chatbots and public health advisories.

Stay tuned for more such updates on Digital Health News