The Ministry of Electronics and Information Technology has selected six companies to build AI copilots, chatbots, and smart assistants for government departments. The move signals a serious investment in digital governance. Yet technology experts say the real obstacle is not the sophistication of the algorithms - it is the state of the data those algorithms must work with.
Government departments have accumulated decades of scattered files, incomplete records, outdated databases, and disconnected information systems. Even the most advanced AI cannot provide accurate answers if the information it receives is inaccurate or poorly organized. As one expert put it, the technology is intelligent, but the information ecosystem it depends on is not.
The chatbot that sends you back to a PDF
Technology experts point to the government's AskSarkar chatbot as a clear example of the problem. When users ask about startup schemes, the chatbot reportedly suggests unrelated programs such as Pradhan Mantri Awas Yojana and Swachh Bharat Mission instead of startup-focused options. When asked about India's AI policy, it often responds with a 170-page PDF and a few website links rather than a straightforward explanation.
For ordinary citizens, the expectation from AI is simple: ask a question and receive a direct answer. Instead, they end up searching through lengthy documents and multiple links on their own. The Railway Ministry's AskDisha chatbot reflects the same limitations. A passenger who asks it to book a train ticket is redirected to the IRCTC website, where they must manually complete the transaction. The entry point has changed, but the user experience remains largely the same.
Five decades of data, no shared language
The AI for Government challenge runs deeper than chatbot performance. Government information in India has evolved over several decades. Some records have been digitized, while many continue to exist only in paper files. Different departments maintain information in different formats, and some databases are updated regularly while others are not. This fragmented environment makes it difficult for AI to assemble a complete picture.
Compounding the problem, departments often operate in silos. Information available with one department is frequently inaccessible to another. Similar work is sometimes repeated by multiple agencies, each maintaining its own independent database. Without data sharing across departments, AI cannot connect information to provide full answers. The six selected companies certainly have the expertise to develop sophisticated systems, but AI is only as good as the information it receives.
Data modernization comes before better chatbots
Experts believe India's biggest AI task is not creating smarter chatbots. The real work is modernizing government data itself - digitizing old records, cleaning outdated information, connecting departmental databases, and ensuring that government data remains continuously updated. Only after these structural issues are addressed can AI deliver the speed and accuracy that citizens expect.
Otherwise, even the most advanced AI systems risk becoming digital assistants that keep pointing users toward another PDF, another website, or another government portal. The pattern is familiar to anyone who has visited a government office in the past: check another file, visit another office, bring one more document. The technology has changed, but the experience often feels exactly the same.
Why this matters for government professionals
For officers and staff inside the system, this analysis carries a practical message. AI tools will arrive in your department, but they will only be as effective as the data you feed them. The six-company selection is a procurement milestone, not a transformation. What happens next depends on the unglamorous work of auditing, cleaning, and connecting the information your department holds.
Understanding how AI interacts with data infrastructure - and why fragmented records produce poor results - is becoming a core competency for public sector professionals. The AI Learning Path for Policy Makers addresses precisely this gap: how to govern information so that technology can actually serve citizens. Without that foundation, even the best-funded AI initiative will produce little more than a smarter redirect to another PDF.
Your membership also unlocks:
