AI in Classrooms: Hanoi Reports 92% Student Engagement — What Leaders Should Fix

Hanoi’s recent pilot programs have pushed AI in education from theory to practice. Early results are eye-catching: internal and independent assessments reported more than 92% student engagement, with notable gains in basic IT skills and lesson retention. The pilots show how thoughtful edtech — devices, modern learning management, curated digital content and teacher certification — can move classrooms quickly. They also reveal the hard work ahead: infrastructure patchiness, uneven teacher readiness and open questions about data trustworthiness and equity.

What the 92% actually measures

Headline numbers invite skepticism, so it’s important to be precise. The 92% figure emerges from a composite of measures used during the pilot at Giảng Võ Secondary and Nghĩa Tân Secondary: activity logs from Google Classroom and Google Workspace for Education, structured teacher observations, and short pre/post assessments for IT skills and targeted learning outcomes (for example, reading-comprehension checks after an AI-supported lesson).

Put simply, “engagement” here meant a mix of digital participation (turning in assignments, active logins), observed classroom participation and measurable skill gains on quick assessments. That blend is useful because no single metric captures learning; taken together, they show strong early adoption in the pilot classrooms. Still, replicability depends on consistency in how those metrics are collected across schools and districts.

Pilot snapshot: what was implemented and why it mattered

Giảng Võ Secondary ran Google Classroom and Google Workspace for Education across three sixth‑grade classes (Literature 6A1, Math 6A6, English 6A9). The school supplied laptops and tablets, trained teachers (many earning Google Certified Educator Level 1), and built a digital repository: 582 electronic lectures at the pilot level and roughly 20,000 school-wide digital resources including 826 e‑books.

Tô Thị Hải Yến, principal of Giảng Võ Secondary School: “The digital classroom is student-centered and integrates technology to encourage active participation, strengthen competencies and character, and provide secure access to digital resources.”

Nghĩa Tân used AI-generated media to support a literature lesson: an AI-created interview with author Nguyễn Nhật Ánh served as a discussion springboard. Teachers treated the AI output as supplemental context rather than a substitute for close reading.

Nguyễn Thị Tuyết, literature teacher at Nghĩa Tân Secondary: “AI should be used as a support tool in literature classes—it can enrich context but does not replace direct reading or human emotional engagement.”

A typical student response captured the classroom impact: a sixth-grader told teachers that having a tablet let them revisit difficult passages at home and try practice problems again — a small detail that helps explain why engagement and retention climbed during the pilot.

Why these pilots worked — and why they might not scale by default

Success in these classrooms boiled down to three coordinated elements: devices and reliable LMS workflows, teacher upskilling, and high-quality digital content. When those align, AI tools and LMS features amplify participation and let teachers move faster toward differentiated instruction.

That said, the pilots also highlight three systemic bottlenecks that threaten scalability:

• Connectivity and devices: Suburban and peripheral districts still report unreliable internet and insufficient devices for every learner.
• Teacher capacity: Earning Level 1 certification proved valuable, but certificates alone don’t create a network of instructional designers. Teachers need ongoing coaching, time for lesson redesign and peer support.
• Data governance and trust: As platforms personalize learning and ingest student data, schools must be able to trace where data came from, who changed it and how it was used.

Nghiêm Hồng Trung, principal of Quốc Oai High School: “Digital transformation in classrooms is a long-term effort requiring comprehensive reforms in governance and teaching organisation; teacher digital skills are foundational.”

Data integrity: blockchain talk and practical alternatives

Policymakers discussing AI in education are rightly focused on data provenance — knowing whether the data feeding AI models can be trusted and audited. Enterprise blockchain (immutable ledger technology) has been floated as one option because it makes records tamper-evident. But blockchain is not a quick-win for every use case.

Practical choices for AI data pipelines in schools include:

• Immutable logs (blockchain or tamper-evident servers): Useful when legal or regulatory audits require unalterable records of who changed what and when. Consider trade-offs around cost and integration complexity.
• Centralized audit servers with strict access controls: Cheaper and easier to integrate, and often sufficient when paired with clear governance and periodic third-party audits.
• Privacy-preserving techniques: Differential privacy and federated learning reduce the amount of sensitive data central systems see while still enabling model improvements.

Choosing the right mix depends on the school system’s risk profile. For routine personalization and classroom management, strong SLAs, encryption, and logged access are typically adequate. For high-stakes analytics or legal evidence, immutable ledgers become more defensible.

Actionable checklist for leaders deploying AI in classrooms

• Define the engagement metric up front: Specify which signals (LMS logs, observation rubrics, assessment scores) count and how they’ll be normalized across schools.
• Invest in two bottlenecks first: Reliable connectivity and one device per learner (or controlled device-sharing + offline sync technologies).
• Move beyond single-course certifications: Build a teacher upskilling pathway that includes lesson redesign hours, peer coaching and an instructional-design role to scale quality.
• Demand data governance from vendors: Require audit logs, data provenance statements, clear retention policies, and parental consent workflows in contracts.
• Evaluate immutability needs: Use tamper-evident logs for financial/administrative data and consider blockchain only where auditability justifies the overhead.
• Measure equity, not just engagement: Track device access, connectivity reliability and learning outcomes by district so pilots don’t widen regional gaps.

Budget and procurement notes for CIOs and procurement teams

Per-student device and connectivity costs vary widely, but procurement should favor total-cost-of-ownership (device lifecycle, warranty, teacher support) and interoperability with Google Classroom and other LMS tools. Explore public-private partnerships for device provisioning and prioritize secure, centrally managed endpoints to limit support burdens on schools.

Where Hanoi goes next

The Hanoi Department of Education plans a broader rollout for the 2026–2027 academic year and is explicit about building a synchronized, data-driven education ecosystem that prioritizes STEM/STEAM, robotics and English proficiency while narrowing regional quality gaps.

Nguyễn Văn Hiền, director of Hanoi’s Department of Education and Training: “Education should cultivate logical thinking, adaptability and creativity so students can master technology; Hanoi aims to build a synchronized digital education ecosystem and narrow quality gaps between regions.”

The pilots show what’s possible when technology, content and teacher capability come together. The next test will be whether Hanoi can close the digital divide, professionalize teacher upskilling beyond badges, and put governance frameworks in place so AI and data work for every student — not just those who are already connected.