Srinivas Narayanan, VP of Engineering, OpenAI
| Photo Credit:
cueapi
India is OpenAI’s second-largest market, Srinivas Narayanan, VP of Engineering said, while underlining the importance of creating indigenous AI models that address culturally specific needs and underrepresented data. He emphasised the need for India-specific AI models to tackle culturally relevant challenges and welcomed innovation in Indic models.
Speaking at IIT Madras Alumni Association’s Sangam 2025 global innovation and alumni summit in Bengaluru, he said, “From OpenAI’s perspective, India is a big user base for us. It is our second largest market. We are interested in solving problems for India. That being said, there should be innovation happening in India, with Indic models solving unique problems for India, and we welcome that. There are use cases that might be more culturally relevant and sensitive. That data may or may not be represented in a larger setting, like if there is proprietary data the country cares about that is not represented in a large model.”
Addressing the debate of whether India needs its own LLM or if SLMs will be sufficient, Narayanan added that both types of models may have a role to play depending on task complexity. While smaller models might be adequate for simpler applications, many real-world use cases demand generalisation and adaptability to entirely new contexts, areas where large models tend to outperform smaller ones.
Narayanan also commented on policy engagement, sharing that the company actively interacts with the Indian government and has a dedicated policy representative based in the country.
“Governments must play a significant role in shaping this technology and deeply understand what they care about, and how to make it useful at citizen scale, what can we use it for. Since this is a new technology, the rate of change is quick. Thinking about policies when the entire landscape is shifting is a new problem,” he said.
He noted that policy should be approached in a way that doesn’t stifle innovation, especially in a rapidly evolving landscape. Premature regulation can hinder progress, so it is important to foster innovation while also ensuring responsible use.
The real challenge, he said, lies in striking that balance. With mature technologies, policy decisions are easier to make because the landscape is more stable. But when things are changing quickly, crafting effective regulation becomes significantly more complex.
Published on July 6, 2025