Page 107 - ISC PROCEEDINGS 21.4

challenge of simultaneously designing, promulgating, and implementing comprehensive
AI governance while maintaining deployment momentum. These challenges demand
proactive, evidence-based policy responses.
5. Policy implications and recommendations
5.1. Completing the institutional and regulatory architecture for AI at scale
Vietnam's first AI Law, effective early 2026, marks a pivotal shift from fragmented AI
pilots to economy-wide deployment. Its risk-based, EU-inspired design balances
innovation with risk governance while emphasizing domestic AI autonomy, aligning with
the Make in Vietnam strategy and the endogenous growth logic of Romer (1990).
Critically, a well-governed institutional environment is a prerequisite for AI to drive TFP
growth: without legal certainty, firms delay the organizational redesign and data
investments that convert AI tools into measurable efficiency gains.
A three-tier institutional framework is recommended. At the national level, the
government should rapidly operationalize the national AI supercomputing center and
open data platform, modeled on Singapore's AI Singapore, South Korea's NIPA, and the
UK's Alan Turing Institute, with secure data-sharing protocols ensuring both public and
private access. At the sectoral level, AI deployment roadmaps for the six priority sectors in
Decision No. 411/QD-TTg (2022) should specify measurable TFP contribution targets,
investment benchmarks, and interoperability standards, directly linking AI investment to
productivity accounting. At the enterprise level, NATIF's mandated AI budget share should
fund SME-focused adoption vouchers for domestically developed solutions, reducing the
fragmentation that prevents AI from scaling beyond pilots into TFP-enhancing
transformation.
Resolution No. 198/2025/QH14 reinforces this framework through tax relief, land
access, innovation ecosystem support, and financing incentives. Combined with purpose-
built AI demonstration zones, Cai Mep Ha (Resolution No. 260/2025/QH), Van Phong, and
Van Don, featuring fast-track regulation and shared AI infrastructure, these mechanisms
create the enabling conditions for AI adoption to generate documented TFP gains scalable
to the national level.
5.2. Human capital strategy: building the complementary assets for AI productivity
GPT theory confirms that human capital complementarity is as decisive as the
technology itself for productivity gains (Brynjolfsson & McAfee, 2014; McKinsey &
Company, 2024). Research consistently shows that AI-driven TFP gains materialize only
when workers can effectively collaborate with AI systems, not merely use them as tools.
In Vietnam, where AI-skilled candidates command a 40% salary premium and over 70% of
the workforce lacks formal qualifications (Public First, 2024), skills development is the
most binding constraint on AI-led TFP growth. A four-track strategy is required.
First, universal AI literacy should be treated as a national public good, echoing the
English-language movement of earlier decades, with the VNeID platform serving as
enrollment infrastructure for a scalable national AI literacy certification system, targeting
at least 90% of the formal workforce by 2030. Second, tertiary education reform must
embed AI, data science, and human-AI collaboration across all disciplines, co-design
curricula with industry under the Law on Science, Technology and Innovation (2025), and
channel part of the 2026 VND 95 trillion science and technology allocation to AI labs and
doctoral fellowships. Third, reskilling the 38 million informal and automation-exposed
workers through AI-enabled personalized learning delivered via digital-government
infrastructure in major cities eliminates attendance barriers at scale. Fourth, international

106