The Promise and the Peril
Hyderabad's NIN is working on an AI-powered search engine and platform to provide nutritional information on food products, a move spurred by the rising consumption of packaged goods and non-communicable diseases like diabetes. For years, Indians have
struggled with global health apps that mistake dal for 'lentil soup' and suggest replacing roti with quinoa. These apps often fail because their databases are built for Western diets, leading to wild inaccuracies and impractical advice. NIN's initiative is a chance to build something that truly understands us. But for this tool to succeed where others have failed, it must pass a series of tough, real-world tests.
1. Can It Speak 'Indian Food' Fluently?
The first and most important test is culinary literacy. The bot cannot simply have entries for 'curry'. It needs to know the difference between a Bengali macher jhol and a Goan fish curry, a Hyderabadi biryani and a Lucknowi one. These aren't just regional variations; they are entirely different dishes with unique ingredients and nutritional profiles. Most existing apps fail this basic test, offering generic labels that render them useless. For NIN's bot to be effective, its AI must be trained on the staggering diversity of our cuisine, from the seven types of sambar in Tamil Nadu to the various saags of Punjab.
2. Does It Understand 'Andaaz'?
Indian home cooking doesn't run on precise cup measurements; it runs on 'andaaz'—that intuitive estimation of spices and ingredients. A mother's 'chammach' of oil is very different from a restaurant's. The bot must be able to handle this ambiguity. Instead of asking for grams, it should understand portion sizes like 'katori' and 'roti'. Existing apps struggle with this, forcing users into a system of measurement that doesn't reflect how they actually cook or eat. A truly smart Indian bot would use visual AI to estimate portions from a photo, accounting for the size of the roti or the depth of the dal in the bowl.
3. Will It Judge My Jalebi?
A nutrition bot that only preaches abstinence from street food and festival sweets is a bot that will be quickly uninstalled. A healthy lifestyle in India isn't about eliminating pani puri or gulab jamun; it's about moderation. The bot's advice needs to be pragmatic. It should be able to say, "You had a samosa for your evening snack, which is high in refined flour and oil. Let's aim for a lighter, protein-rich dinner to balance it out." It needs to work within our cultural context, not against it, providing harm-reduction advice rather than outright prohibitions.
4. Is It Accessible to Everyone?
An AI bot from a national institute should serve the nation, not just English-speaking urbanites. The interface must be available in multiple Indian languages, both in text and ideally through voice commands. Many potential users may not be comfortable navigating complex menus but could easily ask a question in Hindi, Tamil, or Bengali. Accessibility also means not assuming every user has the latest smartphone. The platform should be lightweight and functional even on basic devices with spotty internet connections. It's a public health tool, and its design must reflect that inclusive mission.
5. Is the Advice Actually Practical and Indian?
This might be the most crucial test of all. The bot’s recommendations must be grounded in Indian economic and agricultural reality. It should not be suggesting users eat avocados, kale, or quinoa when affordable, nutritious local alternatives like moringa, amaranth greens, and millets exist. The goal should be to optimise the user's existing dietary pattern, not to replace it with a Westernised ideal. The AI should understand seasonal availability, regional produce, and budget constraints to offer advice that is not just scientifically sound, but also genuinely achievable for the average Indian family.
