Hong Kong's AI grant is generous. Without teacher training, it misses the point.
Hong Kong's AI for Empowering Learning and Teaching Funding Programme gives every publicly funded school up to HK$500,000 for software, hardware and platforms. That's a bold move to modernise classrooms.
But the circular labels "subsidising teachers or parents to enrol/participate in AI-related courses/seminars/workshops" as an improper use of funds. That single line turns a strong idea into a tools-first plan that leaves pedagogy behind.
The core issue: tools don't teach, teachers do
We've seen this movie: interactive whiteboards, tablets, and learning apps that never moved the needle because teaching practices didn't change with them. AI isn't a plug-and-play shortcut; it asks schools to rethink assessment, inquiry, feedback, and ethics across subjects.
Expecting "AI for all subjects" without funded professional development is unrealistic. Give a school a cheque and new devices show up. Give teachers time, training and support and learning actually improves.
What the policy gets right-and what it misses
The grant is generous and the requirement for experience-sharing sessions and open classes is a smart accountability step. But those sessions work best as the culmination of a structured change process, not a starting point.
Right now, the policy funds the vehicle and leaves out the driver. If AI is meant to be the "core driving force behind digital transformation in schools," we need to invest in the people steering it.
What the evidence says
Technology spending alone doesn't improve outcomes. The OECD's analysis of PISA data found countries that invested heavily in education technology saw "no appreciable improvement" in results without strong pedagogy behind it. Source.
How school leaders can make the grant work-now
If the funding can't be used for training, treat procurement as the smaller half of the project and build a parallel plan for people and process.
- Ringfence other budgets: Allocate school-based staff development funds, PTA support, alumni donations or external grants to cover AI training and coaching. Keep it transparent and time-bound.
- Bundle enablement, not courses: Where compliant, negotiate vendor contracts to include onboarding, co-teaching, and in-class coaching tied to product adoption (not off-site "courses"). Confirm terms with EDB before signing.
- Start with 2-3 high-leverage use cases: Examples-feedback on drafts, differentiation for mixed-ability classes, formative assessment item creation, rubrics, and admin workflow automation.
- Build a lead teacher network: Identify 3-5 early adopters per key stage/subject. Give them release time to pilot, document, and model in open classes.
- Redesign assessment: Shift toward performance tasks, oral defences, process portfolios, and transparent criteria. If AI can do the task alone, it's the wrong task.
- Write guardrails: A simple AI use policy covering academic integrity, data privacy, bias, and age-appropriate use. Keep it one page, reviewed each term.
- Coach, don't just train: Pair short workshops with in-class coaching cycles. Teachers change practice through support, not slides.
- Measure impact: Track a few visible metrics-teacher time saved, student revision quality, feedback turnaround, attendance in open classes, student voice on AI use.
Procurement checklist (buy for teaching, not for the brochure)
- Alignment with curriculum goals and assessment needs
- Data privacy, student data handling, admin controls, audit logs
- Chinese and English support; accessibility features
- Interoperability with your LMS/gradebook and existing ID systems
- Total cost of ownership (tokens, seats, training, support)
- On-site support SLAs and clear escalation paths
- Evidence of classroom impact from comparable schools
A 90-day rollout you can actually deliver
- Weeks 1-2: Survey teachers' needs; choose two subjects and one cross-curricular workflow to pilot. Publish success criteria.
- Weeks 3-4: Fast-track procurement for the pilot. Draft your one-page AI use policy. Host a 60-minute staff briefing with live demos.
- Weeks 5-8: Run coaching cycles (plan, teach, reflect). Collect quick wins and barriers weekly.
- Weeks 9-10: Open classes in pilot subjects; capture student/teacher feedback.
- Weeks 11-12: Adjust policy, publish a 2-page playbook, and plan the next subject wave.
For policymakers: one fix unlocks real change
Allow a defined portion of the grant-say 20-30 percent-to fund teacher professional development, coaching, and micro-credentialing tied to classroom practice. Keep the requirements for experience-sharing and open classes, but make them the capstone of a supported change process.
HK$500,000 spent on devices looks impressive on delivery day. HK$500,000 invested in teacher capacity compounds for years.
Two adjacent lessons for school leaders
Governance and early intervention matter. Recent building management failures and subsequent government action show what happens when oversight arrives late. Schools should apply the same discipline: clear roles, transparent reporting, and early escalation when adoption stalls or risks appear.
Community safety and cohesion are non-negotiable. Violent, hateful incidents-wherever they occur-are a reminder to teach anti-bias, digital citizenship, and civil discourse. Create spaces for dialogue, set clear conduct policies, and support students who feel unsafe. Schools can be the steady centre when society feels divided.
If you need structured upskilling
If your grant can't cover training, explore external options that map to classroom practice and ethics. Here are curated, job-specific AI learning paths you can review: AI Learning Path for Teachers, AI Learning Path for Secondary School Teachers, and AI Learning Path for Teaching Assistants.
Further reading
Bottom line: AI can help teachers, but it won't teach for them. Fund the gear, yes-but fund the growth that makes it useful.
Your membership also unlocks:
