Indian government chatbots like Asksarkar return housing schemes when asked about startups. Despite MeitY empanelling six private firms for AI builds, decades of uncoordinated government data makes sophisticated AI almost impossible. Solving the problem requires untangling half a century of bureaucratic mess before any tech solution can work.
How We Got Here
The Ministry of Electronics and IT recently empanelled six private companies to develop AI solutions for various government departments. This follows years of underperforming government AI tools, exemplified by Asksarkar and Askdisha, which consistently provide irrelevant responses or bad user experiences.
The Numbers
- Asksarkar returns Pradhan Mantri Awas Yojana for startup schemes and links to a 170-page PDF or US DARPA Wikipedia for AI policy questions.
- Askdisha, the Ministry of Railways' chatbot, merely directs users to the IRCTC portal for ticket booking instead of facilitating it.
- Both Asksarkar and Askdisha display mobile-width user interfaces even when accessed on a full-desktop screen.
- The underlying issue is not AI capability but 50 years of fragmented and uncoordinated government data practices.
- Lack of inter-ministerial coordination often leads to departments reinventing solutions rather than sharing or building upon existing efforts.
What Happens Next
🇮🇳 Why This Matters for India
For the Bangalore and Delhi-based tech firms empanelled, building for the public sector means navigating a bureaucratic maze before any code gets deployed.
The Take
The focus on "AI copilots" misses the real point: government AI isn't failing due to a lack of LLM power, but because its input data is fundamentally broken. Unless MeitY mandates data standardization across ministries first, these 6 vendors will spend more time on data archaeology than building anything useful.
Source:
The Ken ↗