AI Impact Summit: 'Acceleration of AI diffusion must to benefit all'

At the AI Impact Summit in New Delhi, India unveiled indigenous LLMs and pushed for AI sovereignty as 89 nations backed broader, affordable access to AI

The AI Impact Summit may have seen administrative hiccups and minor disruptions in its early days, but the message from the five-day event is clear — if artificial intelligence (AI) is to benefit all, its diffusion must accelerate, said experts.

 

India used the summit to showcase its sovereign AI ambitions. It announced the launch of four homegrown large language models (LLMs) under the government’s ₹10,372 crore IndiaAI Mission. With this, the country intends to build foundational AI systems that have local context and lower costs.

 

Homegrown startup Sarvam AI, founded by Vivek Raghavan and Pratyush Kumar, launched two indigenous models trained specifically for Indian languages. The move drew praise from Alphabet and Google chief executive officer (CEO) Sundar Pichai.

 

“The work Sarvam has done developing local AI models… I just don’t see any impediments to that, and I think it is very, very well positioned,” Pichai said.

 

Other launches included Gnani.ai’s Vachana TTS, a text-to-speech system capable of voice cloning and generating speech in 12 Indian languages. Also, there is the Indian Institute of Technology Bombay-led consortium BharatGen’s 17-billion-parameter multilingual foundational model.

 

Smaller startups also received attention for applying AI to real-world problems. BigOHealth, for instance, showcased a platform designed to help families of cancer patients better understand next steps based on diagnosis.

 

Commercial success, however, will depend on execution. Still, the optics mark a shift from 2023, when Sam Altman dismissed attempts by startups to compete with OpenAI as “pretty hopeless.”

 

Officials and founders claimed that models launched at the summit can outperform some frontier systems on select metrics.

 

Built at a fraction of the cost of leading global models, these efforts aim to advance the IndiaAI Mission’s objective of building culturally and socially contextualised sovereign AI systems.

 

The New Delhi Declaration, signed by 89 countries as of Sunday, emphasised democratising access to hardware and software resources to ensure affordable access to foundational AI technologies.

 

“The future of AI cannot be decided by a handful of countries or left to the whims of a few billionaires,” United Nations Secretary-General Antonio Guterres said at the event.

 

Experts, including Google DeepMind chief executive Demis Hassabis, computer scientist Yoshua Bengio, and former Infosys CEO Vishal Sikka, discussed what they described as the “jagged intelligence” of AI systems — highly capable at complex tasks yet prone to basic errors. 

"There was a noticeable shift from innovation-first narratives to trust-by-design thinking. Leaders were more candid about the practical realities of scaling AI: from data readiness and cost of compute to security trade-offs and organisational change. This reflected growing maturity, with the focus moving from pilots to performance and from hype to measurable business value," said Nitin Bhatt, technology sector leader, EY India.