Hong Kong tests AI in three departments to speed up services, not cut jobs

Hong Kong picks three departments to pilot AI for faster, better public services

Hong Kong has named the Companies Registry, Transport Department and Buildings Department to pilot AI in administrative work. The goal is clear: improve service speed and quality without increasing costs or cutting jobs.

Led by senior officials with input from private-sector experts, a new AI Efficacy Enhancement Team will steer the rollout across departments. The plan targets AI in 100 procedures in 2026, and at least 200 by the end of 2027, covering data analysis, customer service and the approval of licences and permits.

What changes in the pilots

• Companies Registry: AI will support processing of new company registration applications and annual review reports to reduce turnaround times.
• Transport Department: AI will be applied to high-volume administrative procedures and customer enquiries to ease bottlenecks and improve response times.
• Buildings Department: AI will assist in routine case handling and permit-related workflows to speed up approvals and standardize triage.

Policy stance: efficiency first, headcount stable

• The objective is to enhance efficiency without extra financial expenditure. Cost-cutting is not the driver of this initiative.
• The administration does not equate AI use with job reductions. The aim is to deliver more and better services with existing manpower.
• The Civil Service Bureau will oversee workforce planning, deployment and future recruitment in line with AI adoption.

Governance and timeline at a glance

• AI Efficacy Enhancement Team: Led by senior officials with private-sector experts to advise departments and ensure consistent standards and outcomes.
• 2026: AI integrated into 100 administrative procedures.
• By end of 2027: At least 200 procedures to apply AI across the government.
• Service insights: Hotline data will be analysed to identify common public requests and target high-impact improvements.

What departments can do now

• Map quick wins: List repetitive, rule-based tasks with backlogs (approvals, validations, standard enquiries). Start with low-risk, high-volume use cases.
• Prepare data: Improve data quality, standardize formats, and define retention and access rules. Label small, representative datasets to pilot safely.
• Governance: Set human-in-the-loop checkpoints, decision logs and audit trails. Align with PDPO and guidance from the Privacy Commissioner (PCPD).
• Measure impact: Define baseline KPIs (processing time, error rate, backlog age, customer satisfaction) and track gains every sprint.
• Upskill teams: Train case officers and hotline staff on prompt quality, result verification and exception handling. Create simple playbooks and escalation paths.
• Procurement and security: Use approved tools, set data boundaries, and keep sensitive information off external systems unless cleared.
• Change management: Communicate that AI shifts work from manual repetition to oversight and service quality. Share wins early to build momentum.

What this means for civil servants

• Expect faster case triage, fewer manual checks, and clearer status updates for the public.
• Roles shift toward supervision, exception handling and quality control, not layoffs.
• Better use of hotline data will guide resource allocation and service design.

Build capability

If your team is preparing to pilot AI in admin tasks, consider practical training built around public-sector use cases and data analysis fundamentals. Explore role-focused options here: Courses by job.