Governments must move beyond AI pilots to redesign public services: McKinsey
New Delhi, July 24
Governments need to move beyond isolated artificial intelligence pilot projects and redesign the way public services are delivered to fully harness the technology's potential, according to a McKinsey report.
The report said AI can improve public sector efficiency, enhance service delivery and help governments make better use of public resources. However, it noted that many AI initiatives fail to move beyond the pilot stage due to challenges such as workflow integration, data access, model risks and operating costs.
"Achieving real impact, however, requires more than isolated pilots. Fully realizing AI's potential demands organizations rewire by redesigning processes, ways of working, and operating models to make the technology part of everyday service delivery," the report said.
According to the report, governments should adopt a four-step approach to AI implementation by focusing on mission outcomes instead of technology, redesigning end-to-end workflows, building AI-centric operating models and ensuring human oversight in critical decisions.
The report said the public sector remains behind other industries in AI adoption, with an AI Quotient score of 28 compared with the global average of 33. It attributed this gap to fragmented data, legacy technology systems, workforce constraints and stricter transparency and accountability requirements.
McKinsey said public-sector AI programmes are more successful when they redesign complete service journeys instead of introducing standalone use cases. Around 70 per cent of domain-based AI programmes reach production, compared with 30 per cent of programmes built around individual use cases, it added.
The report also said governments should begin implementing AI alongside efforts to improve data systems rather than waiting for complete data modernisation.
It further recommended that governments focus on measurable outcomes such as reducing waiting times, speeding up public services, improving accuracy and lowering fraud. The report added that for every USD 1 invested in technology, around USD 5 should be allocated to adoption, training and capability building to successfully scale AI across public services.
— ANI
Reader Comments
The AI quotient score of 28 vs 33 global average is not surprising. Our government websites and portals are still clunky, let alone AI integration. But I appreciate the focus on human oversight - we cannot blindly automate services like ration distribution or land records without accountability. Redesign should start with user experience, not technology. 🤔
Finally someone saying what we've been pointing out for years! Indian government has hundreds of AI pilots in agriculture, healthcare, education - but most never scale because they're designed as isolated experiments. The 70% success rate for domain-based programs vs 30% for use-case based is a wake-up call. Time to redesign the entire public service delivery pipeline, not just add AI layer on top of broken systems.
The report's emphasis on measuring outcomes like reducing waiting times and fraud is crucial. I recently applied for a passport renewal - the process took 45 days despite multiple 'AI-powered' tracking features. Meanwhile, Estonia has had AI-integrated e-services for years. We need political will to break bureaucratic silos, not just McKinsey recommendations. 😅
Interesting perspective from an Indian lens. In Australia, we're seeing similar struggles with AI pilots in Centrelink and Medicare. The 5:1 investment ratio for adoption vs tech is actually quite sensible - our Victorian government spent millions on an AI tax system that failed because staff weren't trained properly. India should learn from these global mistakes too.
As someone who works in e-governance
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.