NEW DELHI (India CSR): Infosys Chairman and co-founder Nandan Nilekani recently addressed Indian AI firms, offering crucial advice on the industry’s direction. At Meta’s “Build with AI” summit in Bengaluru, he outlined seven key points, emphasizing the importance of practical applications, efficient data use, and leveraging global models like Meta’s Llama. Here are the main takeaways from his guidance.
1. Prioritize Practical AI Applications Over Large Language Models
Nilekani encouraged Indian AI companies to steer clear of competing with Silicon Valley giants in building new large language models (LLMs). He suggested that India’s focus should be on developing AI applications that solve real-world problems, delivering scalable solutions for sectors like healthcare, agriculture, and education.
“Our goal should not be to build one more LLM. Let the big boys in the (Silicon) Valley do it, spending billions of dollars. We will use it to create synthetic data, build small language models quickly, and train them using appropriate data.” Nilekani advised, urging Indian AI firms to focus on value-driven applications.
2. Develop Robust Data Infrastructure
Data is at the core of AI success, Nilekani highlighted, and creating a strong data infrastructure is essential for India’s AI industry. He believes that the future of AI in India depends on how well companies can organize and utilize data relevant to Indian contexts.
“It’s all about data. How do we create the infrastructure for collecting the right data and make India the use-case capital of AI globally where we actually deploy, add scale and speed in a frugal manner,” he said, emphasizing India’s potential to lead in AI applications globally.
3. Leverage Open-Source Models Like Meta’s Llama
Acknowledging the high costs of LLM development, Nilekani praised Meta’s decision to open-source its foundational Llama models, noting it as a significant opportunity for India. He referred to Llama as a “game changer for us in India and something we need to take full advantage of,” encouraging Indian companies to leverage open-source models to avoid expensive development cycles and focus on adapting these tools to local needs.
4. Embrace Synthetic Data for Tailored Solutions
With models like Llama providing options for synthetic data generation, Nilekani pointed out that Indian AI firms could train smaller, customized models that address specific use cases. This approach allows for efficient and economical AI development, avoiding the need for extensive resources.
“Let other people build LLMs, we will make sure it works for people,” Nilekani emphasized, underscoring his vision of a purpose-driven AI ecosystem that serves people directly.
5. Aim to Make India the ‘Use-Case Capital’ of AI
Nilekani envisions India as the global leader in AI applications. By focusing on practical, use-case-driven AI, he believes India can make an impact in a way that benefits society directly, solving challenges from local communities to national initiatives. Earlier in May, at an event hosted by People+AI, Nilekani articulated this vision, asserting that “The Indian path in AI is different. We are not in the arms race to build the next LLM, let people with capital, let people who want to pedal chips do all that stuff… We are here to make a difference.”
6. Reinforce India’s Unique Path in AI
At the People+AI event, Nilekani outlined a unique path for Indian AI, distinct from global AI races. Rather than striving for AI dominance through model creation, Nilekani emphasized purpose-driven AI that delivers meaningful change, particularly for underserved sectors and communities.
7. Foster Frugal and Scalable AI Solutions
India’s path to AI success, according to Nilekani, lies in building “frugal” solutions that can scale efficiently. This approach maximizes impact without necessitating vast resources, allowing Indian AI companies to create value for a diverse range of local needs while positioning the country as a leader in effective and resource-conscious AI applications.
Meta’s Llama and the Future of Indian AI
Nilekani highlighted Meta’s recent updates to the Llama models, particularly the September release of Llama 3.2, which includes multimodal capabilities enabling it to understand text and images simultaneously. These advancements, along with flexible licensing, offer Indian developers the opportunity to leverage global resources without the burden of developing new models from scratch. With Llama 3.2 available in four variants, Indian AI companies have the flexibility to create diverse applications tailored for India.
You Learn: A Vision for India-Centric AI
Nilekani’s recommendations provide a clear framework for Indian AI firms, focusing on purposeful development, data-driven infrastructure, and scalable solutions. By taking this unique, India-centric approach, the country’s AI industry could set a global standard for socially impactful technology that aligns with real-world needs.
