Artificial intelligence must translate into measurable improvements in public services and citizen welfare, Secretary, Ministry of Electronics and IT (MeitY), S. Krishnan said on the second day of the India AI Impact Summit 2026.
Speaking at a session titled “From Algorithms to Outcomes: Building AI that Works for People,” Krishnan said the India AI Mission has been designed to address diverse needs and real-world challenges rather than remain a purely technological exercise.
“We are providing compute, models and data for one reason only — to build applications with real impact. Whether AI succeeds beyond the hype depends entirely on whether it improves lives,” he said.
He added that AI could significantly enhance productivity in sectors where public systems face capacity shortages. “Governments will never have enough teachers, doctors or judges, but if AI enhances productivity, service quality can improve dramatically,” he said, stressing the need to scale solutions responsibly while protecting privacy and ensuring public money produces measurable outcomes.
Krishnan also urged participants to visit the expo featuring over 600 startups and companies working across healthcare, agriculture, education and manufacturing to see practical applications of AI.
Evidence-based deployment crucial
Iqbal Singh Dhaliwal, Global Executive Director of J-PAL, cautioned against treating AI as a “silver bullet,” stressing the importance of rigorous evaluation.
“We need to ask not just whether AI works, but for whom it works, in what contexts and with what outcomes. Without rigorous evaluation, we risk mistaking excitement for impact,” he said.
Michael Kremer, University Professor in Economics at the University of Chicago, said early evidence shows AI can improve traffic enforcement, automated licensing, healthcare delivery and personalised learning.
However, he noted that adoption within public systems requires careful implementation, procurement reforms and collaboration between governments and philanthropic institutions.
Focus on scalable public applications
The session highlighted that compute capacity, data and models must ultimately lead to deployable solutions that improve governance and service delivery.
Participants agreed that AI’s potential will depend on evidence-based implementation, cross-sector collaboration and responsible scaling, with a focus on citizen outcomes rather than technological experimentation alone.
