India is one of our fastest growing markets: Microsoft AI CEO Suleyman

On his inaugural visit, Suleyman says AI will put knowledge at everyone's fingertips

During his first trip to India, Mustafa Suleyman, Microsoft AI’s chief executive officer, expressed pride that India is one of the company’s fastest-growing markets and highlighted that it has one of its strongest teams worldwide based in Bengaluru and Hyderabad.

 

Suleyman, known for his role in founding DeepMind, one of the world’s leading artificial intelligence (AI) companies, and Inflection AI, shared insights about the future of AI and how it can improve personal well-being.

 

“There are extremely talented engineers and developers here,” said Suleyman at the Microsoft: Building AI Companions for India event in Bengaluru on Wednesday.

 

“We are also increasingly involving social scientists, psychologists, therapists, scriptwriters, and comedians — people you might often associate with the film or gaming industry. This is an opportunity for us to synthesise more diverse perspectives and gain a broader view of those involved in the design and operation process,” he added.

 

 

Suleyman had a fireside chat with S Krishnan, secretary of the Ministry of Electronics and Information Technology, Government of India. When asked about the economic advantages and growth potential that AI can drive in India, especially in a capital-constrained environment, Suleyman said the internet had already put information at everyone’s fingertips.

 

“AI is now going to put knowledge at everyone’s fingertips,” said Suleyman. “This (knowledge) is synthesised, distilled, and personally tuned to the way you want to learn and use information. That applies in the workplace as much as it applies at home.”

 

He gave the example of Microsoft 365 Copilot, an AI-powered productivity tool that helps users complete tasks more efficiently and effectively. It integrates with Microsoft 365 apps like Word, Excel, PowerPoint, Outlook, and Teams to provide real-time assistance.

 

Copilot uses large language models and Microsoft Graph data to provide content and skills that are relevant to a user’s tasks.

 

“It can draw on and provide citations for any question you ask, referencing your email or calendar, looking at your Excel sheets, documents, company human resource records, or supply-chain information,” said Suleyman.

 

This is making a valuable contribution to the workplace. Knowledge workers now have access to useful information on which they can act.

 

“I think that is going to have a profound economic benefit for many of our industries,” he said.

 

The country aims to establish a robust AI computing infrastructure through the India AI mission. When asked about Microsoft’s role in promoting diversity in India, Suleyman said that voice is the ultimate way to make the tools accessible and widely available. He said the government should be making investments in areas such as languages and translation. He also emphasized the importance of access to large government datasets for startups and businesses to train their models and help spur innovation.

 

He highlighted the scientific breakthroughs enabled by AI, mentioning that the 2024 Chemistry Nobel Prize was awarded to John Jumper and Demis Hassabis at Google DeepMind for developing a groundbreaking AI tool for predicting protein structures, called AlphaFold.

 

While discussing the dangers of AI, Suleyman stressed that proactive regulation in the space is crucial. He said it should not be treated as a taboo but as a topic for open dialogue. “Most countries have fairly advanced privacy and security regulations,” said Suleyman.

 

However, he pointed out that the tricky part will be determining when the model (AI) begins to improve itself independently. It is hard to predict how it will evolve, and this may require an interventionist approach. He added that the government’s knowledge and awareness are higher than they have ever been for any other technology.

 

Suleyman also envisioned a new AI-driven experience, where AI acts as a ‘companion’ offering a quieter, softer, and calming digital environment. It adapts to each user’s personal style, goals, and learning preferences.     

Building an LLM from scratch not worth the effort: S Krishnan, Secretary, Ministry of Electronics and Information Technology

 

Microsoft’s Mustafa Suleyman said that pre-trained models would become increasingly commoditised in the future, with many available via API and, in some cases, as open source. He noted that the data required for post-training or fine-tuning a model to adapt it to specific use cases is actually minimal.

 

“You only need a few hundred thousand examples of the behaviour you want the model to imitate or learn from at the post-training stage. I expect to see thousands of agents with various types of expertise, not only linguistically but also in knowledge and grounding across different databases and corpora, emerging in the next few years,” he said.

 

S Krishnan, secretary of the Ministry of Electronics and Information Technology, Government of India, said that during the planning of the ‘India AI Mission,’ there was an initial idea to build India’s own Large Language Model (LLM).

 

“We are now reconsidering whether building an entire LLM from scratch is worth the effort. It may be more effective to adapt existing models to meet India’s specific needs and sectoral requirements,” said Krishnan.

 

Krishnan also shared Prime Minister Narendra Modi’s vision of making AI accessible across India. He highlighted the government’s focus on adapting AI to Indian languages, giving the example of ‘Bhashini,’ an AI-based language translation tool designed to enable real-time translation of Indian languages.

 

In the Indian context, he said there are a few AI-related issues that existing regulations can address, such as concerns over personal data usage, a global issue. Additionally, he noted concerns about AI misuse, including misrepresentation and deep fakes. “I think existing legislation and regulations have been reasonably effective in addressing these issues,” said Krishnan. “The broader question of how we regulate and proceed with AI, given even existential fears, remains open.”