Unlocking AI’s Potential — Without Losing Public Trust!

Artificial intelligence is no longer a horizon technology for government — it’s already deployed in agencies around the world. But a sweeping new study by the Infosys Knowledge Institute reveals a striking gap: public sector organizations are realizing significantly less value from AI than their private-sector peers, and a troublingly high share of projects are being cancelled outright after launch.

45% of public sector AI initiatives achieve their objectives

35% are cancelled or fail to deliver public value

2% of organizations were truly AI-ready entering 2025

The research surveyed 250 executives across 223 public sector divisions — part of a broader study spanning 3,798 business leaders globally. The picture that emerges is one of cautious progress hampered by deep structural friction: legacy systems, fragmented data, regulatory complexity, and an acute shortage of AI-ready talent.

WHERE AI IS GAINING TRACTION

Not all domains are equal. The report identifies a clear divide between AI use cases that are thriving in government and those that remain largely off-limits for now.

Cybersecurity & Resilience Highest viability — detecting threats, protecting critical infrastructure	IT & Operations Automating routine workflows and predictive maintenance
Health & Social Services Eligibility screening, benefits counseling, fraud reduction	Agriculture & Infrastructure Satellite crop monitoring, climate forecasting, asset management

Customer-facing and decision-heavy applications — citizen services, policy recommendations, law enforcement analytics — face much slower adoption. The reasons are familiar to anyone who works in or alongside government: concerns about algorithmic bias, the need for explainability, and the disproportionate public impact if something goes wrong.

THE WORKFORCE GAP HOLDING AGENCIES BACK

Perhaps the most striking finding involves people, not technology. The study categorizes organizations along a maturity spectrum from “watchers” (minimal AI engagement) to “trailblazers” (continuous training and full workforce integration). Public sector agencies are heavily concentrated at the low end — and it costs them dearly: trailblazer organizations are 15 percentage points more likely than explorers to derive measurable value from AI.

The implication is clear. Technical investment alone won’t close the gap. Governance frameworks, data modernization, cross-functional operating models, and sustained workforce development are the differentiators between agencies that experiment and those that transform.

The full report covers sector-by-sector case studies, a detailed use case viability matrix, responsible AI frameworks, and a practical roadmap for public sector leaders. It is essential reading for anyone navigating AI adoption in government.

Read the full report at: infosys.com/iki/research/ai-business-value-radar-public-sector.html