Government Scores 26 Out of 100 on McKinsey's AI Maturity Scale — Here's the Actual Gap

The public sector scores 26 out of 100 on McKinsey’s AI quotient scale, against a 35 global average — and McKinsey’s own research argues the gap has almost nothing to do with technology access. Agencies keep running isolated pilots instead of redesigning how work actually gets done: only 30% of AI programs organized around individual use cases reach production, versus 70% of programs built around whole domains. Roughly 60% of realized AI value, per the report, comes from workflow redesign rather than the underlying model.

The sentiment data is the more uncomfortable number. Only 20% of public sector employees expect AI to meaningfully change their daily work, and just 31% trust their employer to develop it safely. McKinsey’s prescription — mission-driven strategy, end-to-end workflow redesign, capability-building, and preserved human oversight for consequential decisions — comes with a specific spending ratio: roughly $5 on change management for every $1 spent on the technology itself.

That ratio is worth reading against what’s actually shipping into government right now. Anthropic’s public beta of Claude Code and Claude Cowork for government runs inside a FedRAMP High authorized environment, with tamper-evident audit logs built for the Authority to Operate process and department-level spend controls — the compliance and governance layer McKinsey’s framework assumes has to exist before workflow redesign can happen at all. The tooling gap McKinsey describes isn’t really a tooling gap anymore; it’s an adoption-discipline gap, and the infrastructure to close it now exists whether or not agencies are using it.

For consulting engagements advising public-sector or public-sector-adjacent clients, the 26-versus-35 score is less interesting than the $5-to-$1 ratio. It’s a rare case of a research shop putting a real number on how much change management actually costs relative to the AI spend everyone budgets for instead.