Africa's AI Productivity Gain: The Risk of Jobless Growth
An Analysis of the AfDB Report: 'Africa's AI Productivity Gain: Pathways to Labour Efficiency, Economic Growth and Inclusive Transformation' (June 2025)
Kofi Dadzie · January 2026
The AfDB projects 35–40 million net new digital jobs by 2035. But will AI-driven productivity gains actually create jobs, or risk leaving Africa's workers behind?
Introduction
The African Development Bank (AfDB)'s publication 'Africa's AI Productivity Gain: Pathways to Labour Efficiency, Economic Growth and Inclusive Transformation' outlines a strategic roadmap to capture a $1 trillion economic opportunity through artificial intelligence (AI) by 2035.
The Report identifies agriculture, retail, manufacturing, finance and health as target sectors for AI-enabled transformation and growth, potentially creating 35 to 40 million new digital and digitally-enabled jobs. It captures an 'AI-Productivity Flywheel' consisting of five essential enablers: data, compute power, specialized skills, governance trust, and capital, as binding constraints for achieving these gains.
The $1 Trillion Opportunity
Africa can achieve a 24% uplift over baseline GDP projections by 2035, driven by AI adoption across five priority sectors: Agriculture & Food Systems ($200B), Wholesale & Retail ($140B), Manufacturing & Industry 4.0 ($90B), Finance & Inclusion ($80B), Health & Life Sciences ($70B).
This Analysis Focuses On
Whether the 35–40M jobs target is attainable under the Report's assumptions. It addresses attainability vs. gaps, and proposes an actionable research agenda for strengthening the pathway to the goals.
The AfDB Jobs Claim: 35–40M by 2035
What the Report Says
Based on 'Full Flywheel Activation' (Scenario 3), Africa can achieve a US$1 trillion GDP uplift and 35–40 million net new digital and digitally-enabled jobs by 2035. Scenario 3 requires each Regional Economic Community (REC) to run at least one multi-country AI initiative, with 25+ countries exceeding a compute index score of 45/100 (Africa currently ranks 22/100).
The jobs headline is derived via a GDP-to-jobs elasticity factor of 0.4, modeled as:
Net new jobs = 0.4 × 20% GDP uplift × 450M labor force = 36M (rounded to 35–40M)
The Gap in the Evidence
Scenario 3 explains what gets built and activated, but the jobs number is not mechanized through sector-by-sector pathways — no job coefficients per hectare, per clinic, per factory line. It rests on a macro assumption that employment rises with GDP.
The report's employment estimate relies on a fixed GDP-to-jobs elasticity, implicitly assuming that output growth translates proportionally into labor demand. However, a substantial body of task-based research challenges this assumption in the context of AI-driven automation.
Acemoglu & Restrepo — Automation introduces a displacement effect where machines substitute for labor in existing tasks, potentially breaking the historical correlation between GDP growth and employment growth.
IMF on Gen-AI: Artificial Intelligence and the Future of Work — The impact on employment depends on task exposure and complementarity, not aggregate growth alone.
The Core Risk — Employment elasticities should not be treated as structural constants under AI-led growth. Without explicit policy mechanisms, GDP-based job projections risk overstating net employment gains.
A Hybrid Model for Jobs Outcomes
To assess whether the Report's jobs target is attainable, we need a micro-to-meso translation layer: how AI-driven productivity in the target sectors actually yields new work at scale, while also accounting for displacement.
The AfDB model is represented as:
ΔE = εE,Y × ΔY × LF,
where the employment–GDP elasticity is assumed at 0.4, GDP uplift at 20%, and labor force at 450M. The proposed hybrid model replaces this single elasticity factor with a sector-level labor response function.
The Hybrid Model Formula
For each sector s:
ΔEs = ηs·ΔYs − αs·As + βs·Cs + γs·Ns

= (jobs from demand expansion) − (jobs displaced by automation) + (jobs from augmentation) + (jobs from new tasks)"
ηs·ΔYs — Demand Expansion / Scale Effect: Even if AI automates tasks, if it lowers costs and expands output, employment can still rise through volume effects. A low ηs represents the 'jobless growth' phenomenon.
αs·As — Automation Displacement: Acemoglu–Restrepo's displacement effect: tasks previously done by humans are automated. Larger when a sector's tasks have high AI exposure and low complementarity.
βs·Cs — Augmentation / Complementarity: AI makes workers more productive without removing them (copilots, decision support, better diagnostics with humans still acting). Per IMF, this is where AI helps workers do more.
γs·Ns — New Tasks / Reinstatement: Acemoglu–Restrepo's reinstatement effect: the economy creates new tasks, occupations, and services. In Africa, relevant examples could include: agent networks, verification/QA roles, local-language content wrappers, data stewardship.
Policy Recommendations: Steering AI Toward Inclusive Jobs
AfDB's 35–40M jobs target becomes achievable only as a special case where automation displacement is small, augmentation and new tasks roughly offset it, and demand expansion is strong. This requires four clear policy levers.
Augmentation-First Procurement Standards: Government and regulated sectors should procure AI that expands service coverage per worker, not just reduces headcount. Example: AI triage tools that increase nurse/doctor throughput; AI pest detection deployed through extension agents, not replacing them.
Incentives for Labor-Absorbing Service Wrappers: Scale agent, community health worker, and SME-advisor networks that make AI usable in the real economy. Example: AI bookkeeping + compliance service delivered by local agents/accounting SMEs to micro-businesses.
Explicit Transition Rails for Displaced Tasks: Mandatory labor impact plans for large AI deployments, with funded rapid pathways into adjacent roles. Example: Public sector data-entry and form-processing roles staff moved into citizen support, service navigation, local field validation and quality assurance roles .
Market-Expansion Policies in Priority Sectors: Remove bottlenecks so productivity gains lead to more output and more transactions, not merely fewer workers doing the same output. Focus: finance, logistics, standards, exports, reimbursement models.
The policy goal is managed churn with upward mobility, not static job protection.
The Sensitivity Analysis: From Projection to Governance
To operationalize the hybrid model, we translate its parameters into a policy-driven sensitivity analysis anchored on Africa's projected 2035 labor force and the AfDB's assumed AI-driven GDP uplift. The model takes the AfDB's full activation outcome (35–40 million net new jobs) as a reference ceiling, not a default.
Policy levers are applied to test how different implementation choices translate the same GDP growth into sharply different labor outcomes, ranging from job-poor growth to broadly inclusive expansion. The sensitivity analysis is inspired by the ILO’s drive towards “A policy tool for inclusive transitions".
How the Model Works
01
Adjust Policy Levers: Policymakers set global and sector-specific policy settings across four dimensions: augmentation-first procurement, labor-absorbing service wrappers, transition rails, and market-expansion policies. Each lever is rated 0 (none) to 3 (high).
02
Observe Employment Effects: The model decomposes net employment effects into: automation-driven task displacement, augmentation and complementarity effects, new task and service creation, and demand expansion from productivity-led growth.
03
Track Net Jobs vs. AfDB Target: Results are tracked relative to the AfDB's 35–40M target range. Under current baseline settings (partial policy activation), the model would yield net new jobs well below the AfDB target, highlighting the governance gap.
AI-enabled growth is not inherently job-creating or job-destroying — the employment outcome is a function of policy design.

Request access to an interactive version of the sensitivity model to explore sector-specific outcomes and policy trade-offs: tech_policy@kofidadzie.com.
Research Agenda & Next Steps
To strengthen the pathway to the AfDB's jobs goals, the following additional studies and interventions are needed:
1
Sector-Level Jobs Pathways: Build sector pathways for 'jobs created' vs 'jobs displaced' in the five priority sectors, including informal-economy channels, rather than relying on aggregate elasticity.
2
Hybrid Labor Module: Replace the single elasticity with a hybrid labor module that explicitly separates: automation displacement, augmentation/complementarity, and new task creation and demand expansion.
3
Task-Based Labor Disruption Lens: Add a task-based lens consistent with IMF's work on Gen-AI exposure and complementarity to quantify substitution risk and target policy toward high-complementarity adoption pathways.
4
Digitally-Enabled Job Archetypes: Derive 'digitally enabled job archetypes' that AI adoption creates or reshapes (e.g., agents, field workers, micro-entrepreneurs, verifiers, data and compliance roles), and test whether these roles can realistically scale to absorb tens of millions of workers.
AfDB's elasticity-based jobs estimate requires sector modularity and job-preservation/transition policies to increase achievability of the target outcome.
Bibliography