AI Conference 2024: Sandhya Devanathan talks Meta AI in India and how it's aiming to democratise AI for billions

The India Today-Business Today AI Conference 2024, themed “AI for India: Solving Today, Shaping Tomorrow,” hosted an insightful session featuring Sandhya Devanathan, Vice President and Head of India at Meta. Moderated by Aayush Ailawadi, the session delved into Meta’s AI initiatives, its mission to democratise AI, and the transformative role it envisions for artificial intelligence in India.

Meta AI: Empowering Connections Through Technology

Meta AI, Meta’s flagship artificial intelligence initiative, has rapidly gained traction in India since its launch. With over 500 million users globally, India represents the largest user base, spanning urban metros to rural villages. Devanathan emphasised the company’s mission: “We are about human connection and the technology that enables it. Meta AI is central to that mission by enabling and democratising access to AI for billions of people.”

From generating responses on Meta AI to assisting users with day-to-day tasks, the platform has seen diverse applications. Devanathan shared how users in India are leveraging Meta AI, citing examples such as a mother using it to help her child with homework and another user creating a birthday card for a loved one.

“What we’re seeing in India is unique. Indians are innovating on how they use Meta AI, from simple everyday tasks to more advanced creative endeavours,” she said.

Scaling AI Responsibly with Open Source Collaboration

Meta’s philosophy of open-sourcing its technology has been pivotal in fostering innovation. Devanathan highlighted the open-source nature of Meta’s LLaMA models, which have been adopted by various developers and enterprises in India. “We believe open-sourcing AI technology builds trust, scale, and robustness. It enables a whole community of developers to collaborate and improve these systems,” she noted.

Indian organisations are already creating solutions tailored to local needs using Meta’s AI models. For instance, Sikka.ai developed India’s first Hindi language model using LLaMA, while Odia-gen focuses on audio-based AI solutions. These innovations are enabling solutions across India’s diverse linguistic landscape, which includes 22 official languages and over 700 dialects.

Privacy and Responsibility: A Core Focus

Addressing concerns about data privacy and security, Devanathan emphasised Meta’s commitment to building responsible AI systems. “We take privacy seriously. Every product undergoes extensive privacy reviews before launch. For example, our Meta Ray-Ban glasses have safeguards like a visible light indicator to ensure privacy is maintained,” she said.

Meta’s approach to responsible AI development includes rigorous internal and external testing. Devanathan explained the company’s “red teaming” initiative, which simulates potential threats to AI systems to identify and mitigate vulnerabilities. “We want to build systems that people trust and find useful. It’s an ongoing effort, but one we take very seriously,” she added.

The Future of AI in India

Devanathan believes that AI has immense potential to drive growth and innovation in India, particularly through affordable and accessible solutions. Meta’s collaborations with governments and enterprises aim to enhance education, skilling, and citizen services. She pointed out the company’s work with state governments like Karnataka, Andhra Pradesh, and Telangana, and the establishment of AI education centres, including one with IIT Jodhpur.

“India is a key market for Meta AI. Our focus is on empowering users with tools that are intuitive, accessible, and impactful,” she concluded.