policies-for-social-and-economic-resilience-in-the-ai-age-global-south-perspectives

Abstract

The panel examined how governments, NGOs, and private‑sector actors in the Global South can design and scale digital‑first social protection systems that are resilient to AI‑driven economic shocks. Panelists presented concrete case studies—from Togo’s rapid digital cash‑transfer rollout during COVID‑19 to India’s Digital Empowerment Foundation’s “information‑first” model—followed by research‑based policy recommendations on AI‑enabled job creation, cash‑transfer design, and multi‑stakeholder governance. The discussion closed with rapid‑fire policy tips aimed at helping Global‑South policymakers build inclusive, data‑rich, and bottom‑up AI ecosystems.

Detailed Summary

The moderator framed the session around two intertwined challenges: (a) the need for institutional capacity to deliver rapid social assistance, and (b) the risk that AI‑driven productivity gains may concentrate wealth unless coupled with inclusive policy tools.
Emphasis was placed on moving the debate from “future of work” to “future of income” for vulnerable populations.

2. Case Study – Togo’s Pandemic‑Era Digital Cash Transfer

Speaker: Kô Goma (Togo Ministry of Public Service Efficiency and Digital Transformation)

2.1 Problem & Initial Set‑up

Togo lacked an up‑to‑date social registry at the start of COVID‑19.
Voter‑ID numbers were repurposed as parametric identifiers to reach the poorest.

2 .2 Digital Platform Development

Built a USSD‑based payment platform within days, enabling direct transfers to ≈ 920 000 people (≈ 25 % of the national population).
Integrated satellite‑imagery mapping and telecom consumption data to build two machine‑learning models:
1. Asset‑mapping model – ranked districts from poorest to richest.
2. Consumption‑pattern model – identified households with daily expenditures < $2.

2.3 Targeting & Outreach

The ML‑driven list was contacted via USSD; households self‑applied for aid.
Post‑implementation assessment showed the ML‑selected cohort was significantly poorer than the pre‑existing file, confirming higher targeting efficiency.

2.4 Scaling & Institutionalisation

World Bank funding enabled a national Social Resilience Programme:
- Bamitri‑KD – a large‑scale digital registration effort covering the entire Togolese population.
- A dynamic social registry that updates continuously, feeding the same payment platform for rapid response.
Partnership with University of Berkeley to create an in‑house data lab, reducing reliance on external expertise.

2.5 Key Insight

“A crisis‑driven digital cash‑transfer system can become the backbone of a permanent, adaptable social protection architecture if built on open data, local talent, and AI‑enhanced targeting.”

3. Digital Public Infrastructure for Rural Empowerment (India)

Speaker: Osama Manzar (Digital Empowerment Foundation – DEF)

3.1 Core Premise – Information as the First Layer

In villages, trusted, localized information (health, market prices, entitlements) is the foundation for any digital service.

3.2 Human‑in‑the‑Loop Trust Model

DEF operates “Info‑Meri” kiosks in ~3 000 locations, staffed by community members who verify transactions and provide assistance.
The kiosks serve as both payment points and information hubs, creating a two‑way flow between citizens and services.

3.3 Design Principles for Inclusive Digital Infrastructure

Principle	Explanation
Human‑loop of trust	Digital tools must be mediated by known, local actors to build confidence.
Bidirectional exchange	Citizens should also be able to sell products, share knowledge, and receive feedback—not just consume services.
Frugal, low‑cost tech	Solutions must run on basic hardware and cheap data plans; sophistication is secondary to accessibility.
Ethical, non‑biased AI	AI‑driven content must be transparent, unbiased, and not reinforce existing power imbalances.
Bottom‑up design	Systems should be co‑created with communities, not imposed top‑down from Silicon Valley.

3.4 Takeaway

“Digital public infrastructure succeeds when trust, reciprocity, and affordability are woven into its design from day one.”

4. Cash Transfers as an Income‑Floor Mechanism

Speaker: Adrian Brown (Windfall Trust) – representing GiveDirectly

4.1 Organizational Background

GiveDirectly is the world’s largest unconditional cash‑transfer (UCT) organization, having moved ≈ $1 bn to ultra‑poor households across Africa, Asia, and the Pacific.

4.2 Policy Framing – From “Future of Work” to “Future of Income”

UCTs are the most evidence‑backed poverty‑alleviation tool (World Bank, multiple RCTs).
Large, lump‑sum transfers (as opposed to recurring payments) have shown stronger impacts on entrepreneurship, migration decisions, and asset accumulation.

4.3 Nuances in Cash‑Transfer Design

Targeting trade‑offs:
- Universal vs. selective: universal approaches avoid exclusion errors but cost more.
- Lump‑sum vs. recurring: lump‑sum enables large‑scale investment; recurring stabilises consumption.
Implementation considerations:
- Leverage existing private‑sector payment rails (e.g., India’s UPI).
- Embed digital identity (voter IDs, mobile numbers) for rapid disbursement.

4.4 AI‑Enabled Redistribution

Cash transfers can counteract wealth concentration that AI may generate by providing a floor for the most vulnerable, allowing them to adapt (e.g., reskilling, small‑business start‑ups).

4.5 Call to Action

Policymakers should institutionalise UCTs as a core component of AI‑era social safety nets, backed by robust data and transparent delivery channels.

5. AI‑Driven Job Creation – BCG & NITI Aayog Report

Speaker: Sambhav Jain (BCG)

5.1 Research Scope

Co‑authored “Roadmap for Job Creation in the AI Economy” (BCG + NITI Aayog).
Focused initially on the technology sector, but framework applies across industries.

5.2 Quantitative Findings

Metric	Finding
Productivity gains (current)	15‑20 % in tech firms.
Projected productivity (2030)	30‑40 % – still far from hype of 50‑80 %.
Jobs disrupted (WEF)	60‑70 million.
Jobs created by AI (WEF)	170 million (net gain).

5.3 Emerging AI Roles (Three “W” categories)

Enterprise AI – AI product managers, AI Ops engineers, Prompt engineers.
Frontier AI – AI‑Cybersecurity, AI‑Haptics, AI‑Quantum specialists (fast‑growing).
AI‑for‑AI – Researchers building next‑generation agentic AI (still niche, Western‑biased).

5.4 South‑South Challenges (India as exemplar)

Challenge	Description
Skill‑stock gap	Existing engineers lack AI‑centric upskilling.
Talent‑flow gap	Universities produce few AI PhDs; brain‑drain to global hubs.
Supply‑demand mismatch	Companies struggle to find qualified AI talent, leading to loss of potential jobs.

5.5 Five‑Fold Policy Recommendations

Talent Magnetism – Visa/visa‑type programmes to attract global AI talent.
Curriculum Revamp – Nationwide integration of AI modules from primary to postgraduate levels.
Industry‑Academia Partnerships – Co‑create reskilling bootcamps and apprenticeships.
Compute Infrastructure – Build affordable, high‑performance compute clusters accessible to local researchers.
Open‑Data Ecosystem – Publish curated, anonymised datasets for AI model training.

5.6 Core Message

“The Global South must capture a fair share of AI‑generated jobs by closing talent pipelines, democratizing compute, and fostering open data.”

6. Multi‑Stakeholder Governance for Shared Prosperity

Speaker: Rebecca Finlay (Partnership on AI)

6.1 Partnership on AI (PAI) Scope

140 + organisations across 18 countries collaborate on responsible AI.

6.2 “Future of Work” as a Convening Theme

PAI identified future of work as the most pressing responsible‑AI challenge, demanding cross‑sector dialogue.

6.3 Shared‑Prosperity Guidelines (SPG)

Framework for Job Impact Assessments (JIA) – systematic evaluation of AI adoption on different occupational groups.
Spaces for Worker‑Centred Innovation – co‑design labs where employers, labour unions, and policymakers prototype AI‑enabled workflow redesigns.

6.4 Current Activities

Piloting use‑case prototypes in real‑world settings (e.g., AI‑assisted scheduling in manufacturing, AI‑driven upskilling platforms).

6.5 Policy Implication

“Multi‑stakeholder processes must be worker‑led from the start, ensuring AI benefits are measured, disclosed, and redistributed through concrete policy levers (e.g., tax‑rebates, UCTs).”

Speaker: Thomas Schneider (Swiss Federal Office of Communications)

7.1 Cultural Foundations

Switzerland’s mountain‑community heritage forged a bottom‑up, mutual‑aid culture: neighbors cooperate during avalanches or political threats.

7.2 Institutional Manifestations

All‑party coalition government – social consensus across linguistic (German/French/Italian) and political lines.
Grass‑roots labour movements historically secured fair shares of productivity gains, preventing extreme inequality during industrialisation.

7.3 Lessons for the Global South

Self‑organisation – resilient societies rely on local coordination rather than top‑down directives.
Inclusive dialogue – multi‑language, multi‑cultural negotiation ensures policies are broadly acceptable.
Scenario‑based planning – develop global‑level simulations to anticipate AI’s impact and negotiate collective responses.

7.4 Call to Action

“Build strong, inclusive coalitions that can rapidly respond to AI‑driven disruptions, anchored in transparent data, trusted institutions, and shared decision‑making.”

8. Quick‑Fire Policy Recommendations (All Panelists)

Speaker	One Concrete Recommendation for Global‑South Policymakers
Adrian Brown	Collaborate across borders – share platforms, data, and AI expertise (e.g., the “Novissi” model).
Stella Luk	Invest in robust payment pipelines (USSD, UPI‑style systems) and augment them with AI‑driven shock‑response tools.
Sambhav Jain	Launch a bottom‑up skills‑to‑jobs initiative focusing on the five biggest employment sectors, defining future skill‑sets and creating local demand incentives.
Rebecca Finlay	Prioritise multi‑stakeholder early‑warning systems that provide real‑time data and involve workers from the outset.
Thomas Schneider	Scale education and fight corruption – ensure talent can thrive regardless of background, and that meritocratic pathways are protected.

9. Closing Remarks (Moderator)

The moderator thanked the panel and announced the next session, distributing mementos and asking the audience to exit the hall.

Key Takeaways

Digital cash‑transfer platforms can be built in weeks when existing identifiers (e.g., voter IDs) and low‑tech USSD interfaces are leveraged.
AI‑enhanced targeting dramatically improves poverty‑reach, but must be paired with transparent, locally‑managed data pipelines.
Information‑first, trust‑centric design is essential for any digital public infrastructure serving rural or informal communities.
Unconditional cash transfers remain the strongest, evidence‑based tool for building an income floor that enables resilience and entrepreneurship amid AI‑driven economic shifts.
AI will create more jobs than it destroys (≈ 170 M new vs. 60‑70 M displaced globally), yet the Global South faces talent‑stock, flow, and supply‑demand gaps that risk marginalising it from AI‑generated opportunities.
Five strategic levers (talent attraction, curriculum overhaul, industry‑academia reskilling, compute access, open data) are required to capture AI’s job‑creation potential.
Multi‑stakeholder governance—bringing together governments, private firms, academia, and civil society—is critical for transparent impact assessments and for shaping AI policies that prioritize human welfare.
Bottom‑up, community‑driven models, exemplified by Swiss cultural practices, provide a scalable blueprint for collective resilience in the face of AI‑induced disruption.
Collaboration, payment‑infrastructure, inclusive skill development, data‑driven early warnings, and anti‑corruption measures are the concrete policy pillars repeatedly highlighted across the panel.

See Also:

India AI Impact Summit 2026

Explorer