FGI LIVE Q3: Human Experience — AI that measurably improves services
About Event:
Every government now has an AI pilot. Few can point to a service that got measurably better: faster resolution, less failure demand, fewer complaints, wider access. Closing that gap is the agenda.
The provocation we open with: one of the world's most advanced government AI delivery programmes has published its full playbook openly, and it still keeps AI agents away from citizen-facing services. Its own papers say the trust boundary is not ready. If the furthest-ahead practitioners hold that line, a chatbot on the website is not a strategy. So what does real, measurable AI-enabled service improvement look like, and what must be true first?
Executive focus: AI is only real if it reduces friction and failure demand, and only if the underlying service is already at Gov 3.0 readiness.
Readiness lens: We open with a 5-minute Gov 3.0 readiness check against the FGX Human Experience indicators, so you leave knowing what must be true in your jurisdiction before AI goes anywhere near a citizen-facing service.
What we'll work through:
Where AI is improving services in practice (triage, casework, contact centres, eligibility) and where it just moves the queue.
The foundations that must be green first: a named owner for each end-to-end service, service data you can trust, and a human fallback built as permanent infrastructure.
Inclusion as a design constraint: content, language access, and the digital inclusion floor.
Measurement that survives scrutiny: time saved, failure demand reduced, complaints, inclusion and access. The working test: a measure that cannot show failure cannot show success.
You leave with: a readiness read for your own jurisdiction, patterns from peers who have shipped, the session Insights Brief, and the quarter's working tool.
FGI Live is designed for government executives and senior leaders, digital transformation practitioners, policy makers, and members of the Future Government Institute community. This is a working roundtable. Nothing shared is attributed to any individual or organisation.