Vietnam's digital shift is remaking work: platform and AI-mediated jobs increase flexibility but shift income risk, surveillance and control onto workers, exposing gaps in social protection and worker voice; trade unions must broaden strategies across representation, rights, reskilling, reach and social dialogue to safeguard decent work.

Main Finding

Digital transformation in Viet Nam is restructuring work not primarily by eliminating jobs en masse but by transforming tasks, control mechanisms, skills, and representation. An estimated 11.5 million workers (~1 in 5) perform occupations with tasks potentially exposed to generative AI; fewer than 2% of workers are in occupations combining high AI exposure with a high potential for full automation. These changes create mixed opportunities (flexibility, new tasks, productivity gains) and risks (income insecurity, skills gaps, work intensification, data surveillance, weakened voice). Trade unions and labour institutions must adapt across five domains—Representation, Rights & Protection, Reskilling, Reach, and Relations & Dialogue (the proposed 5R framework)—to preserve decent and sustainable work.

Key Points

• Four interconnected shifts identified:
  • From standard employment toward diverse, digitally mediated forms of work (digitally augmented, remote/hybrid, platform-mediated, AI/algorithm-mediated).
  • From direct managerial supervision toward algorithmic/digital forms of control (algorithmic management using ratings, automated task allocation, surveillance).
  • From relatively stable occupational skills toward continuous digital competence and adaptability.
  • From workplace-centred worker representation toward the need to represent dispersed, flexible, platform-connected workers.
• Platform work exemplifies blurred boundaries: platform architectures create multi-actor relationships (Platform → Algorithm → Worker → Customer) where algorithmic systems often exercise de facto managerial control even when workers are formally self-employed.
• Empirical measurement challenges: the 2023 pilot DPE module in Viet Nam’s Labour Force Survey (Hanoi & Phu Tho) highlighted sampling, reference-period, and misclassification difficulties—underscoring the need for dedicated statistical instruments for platform employment.
• Distributional outcomes are heterogeneous: AI tends to transform tasks (complementarity for some, substitution for others), producing potential productivity gains but also risks of greater inequality due to uneven digital skills and gaps in social protection.
• Worker voice and collective bargaining face new obstacles when managerial authority is technologically mediated or when workers are geographically/dispersedly connected via platforms.
• Policy and institutional response recommended via the 5R framework: Representation; Rights & Protection (including social protection and legal status); Reskilling; Reach (extending services to dispersed workers); Relations & Dialogue (social dialogue over algorithmic governance and data policies).

Data & Methods

• Study design: qualitative synthesis using structured document analysis and secondary data analysis; not a systematic review but a theoretically guided thematic synthesis.
• Source categories:
  • Peer-reviewed literature (primarily 2020–2026, indexed in Scopus/Web of Science).
  • ILO reports and working papers (used for occupational exposure indices, DPE measurement guidance, and international comparisons).
  • Official Vietnamese sources (labour-force statistics, legislation, policy).
  • Practice-based media/trade-union materials (Báo Lao Động; Lao động & Đoàn thể) for emergent concerns and responses.
• Key empirical datapoints cited in the article:
  • ILO-based application to Viet Nam’s 2024 Labour Force Survey: ~11.5 million workers in occupations with tasks potentially exposed to generative AI (Weidenkaff & Huynh, 2026).
  • Pilot digital platform employment (DPE) module added to the LFS in Dec 2023 (Hanoi & Phu Tho) with ILO technical support—used to demonstrate measurement challenges.
  • Finding that fewer than 2% of workers are in occupations with both high AI exposure and high automation potential.
• Analytical approach: thematic content analysis across five themes (T1 employment transformation; T2 skills/adaptability; T3 working conditions/outcomes; T4 labour relations/voice; T5 trade unions/social dialogue).

Implications for AI Economics

• Task-level transformation dominant: AI’s primary economic effect in Viet Nam appears to be task reallocation and augmentation rather than large-scale job destruction. Economic models should therefore emphasize task complementarities/substitution at the task (not job) level and incorporate on-the-job reallocation of time and effort.
• Labor supply and wage effects will be heterogeneous: occupations facing task augmentation may see productivity and wage gains for workers with complementary skills, while other groups may face downward pressure, precarity, or increased non-wage risks (e.g., loss of bargaining power). Distributional analysis and inequality metrics should be central.
• Algorithmic management changes bargaining power: opaque, automated allocation and evaluation can shift economic risk onto workers and erode traditional levers of worker power. Empirical work is needed to quantify how algorithmic control affects wage-setting, outside options, and collective bargaining outcomes.
• Measurement and identification challenges: existing labour statistics undercount or misclassify platform-mediated and algorithm-mediated work. AI-economics research needs improved measurement frameworks (task modules, platform modules, reference-period design) to correctly estimate employment, hours, earnings, and job transitions.
• Social protection and compensation design: the prevalence of platform-like arrangements implies a need to rethink safety nets (earnings smoothing, portable benefits, unemployment insurance adapted to intermittent work). Cost–benefit analyses of such reforms should be prioritized.
• Skills-policy implications: investment in reskilling and lifelong learning has direct economic returns by shifting workers into complementary roles with AI. Evaluations should consider heterogeneity in access to training and returns across sectors and demographics.
• Research priorities suggested by the paper:
  • Quantitative task-level studies estimating complementarities/substitution between AI and human tasks in Viet Nam sectors.
  • Causal analysis of algorithmic management’s impact on worker earnings, hours, and turnover.
  • Measurement research improving DPE identification in national surveys (sampling strategy, reference periods, classification).
  • Welfare analyses of policy interventions (portable benefits, training subsidies, regulation of algorithmic transparency).
  • Empirical studies of how union/collective responses (or their absence) mediate the economic effects of digitalisation on wages and job quality.

Suggested immediate steps for economists studying AI in labour markets in Viet Nam: - Use task-based exposure indices (as in the ILO application) combined with administrative and platform data to map likely winners and losers. - Incorporate algorithmic management variables (surveillance intensity, automated allocation metrics) into labor supply and bargaining models. - Coordinate with national statistical offices to pilot improved DPE modules and collect richer task-level data.