Algorithmic oversight in public transport—from constant GPS tracking to automated deactivations—correlates with higher anxiety, burnout and chronic precarity among drivers; authors call for transparency, mandatory mental-health risk audits, human review of sanctions and wage protections to curb psychological harm.

Main Finding

Algorithmic management in public transport and logistics acts as a form of psychological governance: automated direction, evaluation and sanctioning (GPS/telematics, ratings, dynamic pricing, deactivation) reshape work design and raise measurable occupational mental-health risks (burnout, anxiety, depressive symptoms, hyper-vigilance, chronic precarity). The authors synthesize evidence into a three‑level (technological → organizational → psychological) framework and propose specific pathways linking technical features to mental-health outcomes.

Key Points

• Scope and synthesis: PRISMA-guided systematic integrative review of 48 empirical studies (2016–2025) drawn from 4,812 initial records across Scopus, Web of Science, PubMed and institutional reports.
• Core algorithmic mechanisms identified: GPS tracking (continuous surveillance), user rating systems (emotional labour and social evaluation), dynamic pricing (income volatility), automated sanctions/deactivation (fear and insecurity).
• Quantitative patterns from included studies:
  • 59.6% of platform workers face “digital speed determination” (higher time pressure).
  • 36.3% more platform workers are subject to third‑party ratings versus traditional workers.
  • High prevalence ranges reported: depressive symptoms 23–41%; high burnout (emotional exhaustion) 41–67%; 78% report chronic fear of deactivation in some samples.
  • Reported effect sizes/propositions include surveillance intensity → reduced wellbeing/hyper‑vigilance (β = -4.213) and algorithmic completeness negatively predicting autonomy (r ≈ -0.42).
• Thematic propositions developed: surveillance → hyper‑vigilance; opacity → procedural anxiety/“algorithmic paranoia”; income volatility → depressive symptoms; rating pressure → emotional exhaustion; task fragmentation → reduced sense of accomplishment; deactivation fear → chronic precarity.
• Policy recommendations emphasized by authors: algorithmic transparency mandates, mandatory mental‑health risk audits, participatory co‑design of systems, human review of deactivations, and minimum‑wage protections in line with ILO principles.
• Limitations in the literature: many studies are cross‑sectional, geographically clustered, short‑term, and heterogeneous in measures—causal inference and long‑run health impacts remain underexplored.

Data & Methods

• Review design: systematic integrative review following PRISMA 2020 to combine qualitative, quantitative and mixed‑methods evidence and integrate psychological theory with technical features.
• Information sources: Scopus, Web of Science, PubMed; additional institutional reports (ILO, EU‑OSHA); backward citation searching.
• Screening/results: 4,812 records identified → 3,120 after de‑duplication/screening → 481 full texts assessed → 48 studies included.
• Inclusion criteria: peer‑reviewed empirical studies (qualitative, quantitative, mixed) addressing automated managerial functions (assignment, evaluation, discipline) in transport/logistics and reporting occupational mental‑health outcomes (burnout, anxiety, stress, depressive symptoms).
• Data extraction: structured template capturing study metadata (author, year, country), sector, sample size/design, algorithmic mechanism(s), psychological outcomes, mediators/moderators (perceived fairness, autonomy, algorithmic literacy).
• Synthesis: six‑phase thematic analysis using Braun & Clarke approach, organized at three levels — Technological (data collection, ML models), Organizational (work allocation, evaluation, discipline), Psychological (rumination, ego depletion, detachment).
• Trustworthiness: triangulation across methods and settings; peer debriefing; emphasis on multilevel propositions grounded in extracted evidence.

Implications for AI Economics

• Labor supply & participation: algorithmic governance alters job quality (loss of autonomy, higher instability). This can lower retention, change reservation wages, and alter occupational choice—potentially reducing labor supply in affected segments unless compensated by higher nominal pay or flexibility value.
• Wage dynamics & income risk: dynamic pricing and automated task allocation increase income volatility and precarity. Economically, this raises the effective risk premium workers require and may increase demand for wage smoothing mechanisms (minimum guarantees, wage floors, or insurance).
• Bargaining power & market structure: algorithmic opacity and automated deactivation weaken worker bargaining and increase monopsony‑like power for platforms that control assignment, evaluation, and sanctioning. This may depress equilibrium wages and increase asymmetric information rents captured by platforms.
• Productivity vs. externalized costs: while algorithms can increase operational efficiency (matching, routing, utilization), a portion of productivity gains may be externalized onto workers via higher intensity, stress, and health costs. Ignoring these externalities risks overestimating net social gains from platformization.
• Human capital and long‑run supply: chronic psychological strain (burnout, reduced accomplishment) can degrade human capital accumulation (skill depreciation, early exit), affecting long‑run labor quality and raising replacement/training costs.
• Policy and regulation implications for economists/policymakers:
  • Transparency and auditability: regulations requiring explainability, reporting of algorithmic effects on work design, and independent audits could reduce information asymmetries and correct market power distortions.
  • Income guarantees and social insurance: subsidies, minimum pay rules, or mandated earnings floors can mitigate income volatility externalities and support labor supply.
  • Enforcement of human oversight: mandatory human review for deactivations and appeal mechanisms protect against automated errors and reduce psychological harm, which otherwise translates into productivity and turnover costs.
  • Measurement & evaluation: incorporate psychosocial risk indicators into cost‑benefit analyses of platform deployment; include mental‑health externalities when estimating social returns to algorithmic adoption.
• Research gaps of economic relevance: need for longitudinal causal studies to quantify impacts on wages, labor supply elasticity, turnover costs, productivity net of health externalities, and distributional effects across Global North/South contexts. Better measurement will inform optimal regulatory design (tax/subsidy or direct regulation) and the appropriate scope of labor protections.

If you want, I can convert these implications into specific policy levers (tax/subsidy models, regulation templates, or sketch an empirical research design to estimate the wage impact of algorithmic surveillance).