## How can artificial intelligence benefit more people in developing economies?

## Source details

**Canonical URL:** [How can artificial intelligence benefit more people in developing economies?](https://www.imf.org/-/media/files/publications/fandd/article/2026/09/pov-tan.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/fandd/article/2026/09/pov-tan.pdf.md)
- [Structured JSON version](/-/media/files/publications/fandd/article/2026/09/pov-tan.pdf.json)

---

### Platform power
- Southeast Asia context: "Here, 78 percent of the workforce, numbering hundreds of millions, operates in the informal economy."
- Platforms can extend financial access:
  - "Last year Grab disbursed $3.2 billion in loans to driver partners and merchant-partner borrowers."
  - "Sixty-eight percent of those borrowers accessed formal credit for the first time through our platform."
  - Platform credit design features: AI reads platform-generated records of earnings and activity to judge creditworthiness; credit limits set based on platform earnings; repayments deducted in proportion to daily income; hardship relief granted case-by-case if earnings cease.
  - "Half of driver-partner borrowers say they turned to Grab to avoid unlicensed lenders."
- Platforms improve earnings and matching:
  - "Average driver-partner earnings per online hour rose by 5.5 percent year over year last year."
  - Platform capabilities: assess real-time demand, match driver partners, predict best spot for next job, shrink unpaid gaps between jobs.
- Real-time behavioral data substitute where documentation fails; platform-based distribution reaches where institutional access does not.

### Skills that pay
- Investment in AI literacy and practical skills is essential for informal workers to benefit from AI-enabled tools.
- Training reach and impact:
  - "More than 1.5 million driver partners completed at least one GrabAcademy course last year."
  - "One in four micro, small, and medium enterprises on the platform now uses Grab’s Merchant AI Assistant."
  - Use of Merchant AI Assistant "underpinned a 15 percent lift in sales."
- Challenge: informal workers lack formal employers to provide training on paid time and at a fixed workplace; public-private partnerships can scale codesigned, flexible training.

### My three asks (policy recommendations)
- First: "fund AI literacy as a public good."
  - Treat AI literacy funding with the same seriousness as roads or energy.
  - Scale permanent, codesigned training with local institutions, structured for workers with irregular hours.
  - Example: SkillsFuture Singapore recognized Grab as an industry leader in championing skills development.
- Second: "rethink how success is measured."
  - Governance frameworks should adopt metrics suited to the informal economy: earnings stability, earnings volatility, participation in multiple platforms, access to credit, breadth of participation.
  - Avoid metrics calibrated solely around formalization rates or financial stability that miss informal-economy priorities.
- Third: "unify regional AI governance."
  - Diverse regulatory frameworks across Southeast Asia constrain platform decisions on data storage and scaling.
  - The May 2026 Association of Southeast Asian Nations’ Digital Economy Framework Agreement establishes regional coordination architecture.
  - Policymakers must ensure the framework "works for the informal economy and does not impose formal economic assumptions on a regional scale."

### Designing for informality (broader lessons and protections)
- The informal economy is a structural foundation and its flexibility and adaptability are features, not failures.
- Regions where informal work is predominant:
  - "informal work accounts for 86 percent of employment in Africa, 68 percent in Asia, and 51 percent in Latin America."
- Policy design principle: build measures around informality rather than importing systems developed for different economic conditions.
- Protections should travel with the worker:
  - Portable social insurance as an alternative to formal-employee classification.
  - Example: "Malaysia’s March 2026 Gig Workers Act... extending social security coverage to gig workers through the national program without making them employees."
- Collaboration between governments and platforms is a consequential policy investment to close the AI divide and scale benefits beyond a single company.

*anthony tan is chief executive and cofounder of Grab; Point of View, F&D, SEPTEMBER 2026.*

---


_Source: https://www.imf.org/-/media/files/publications/fandd/article/2026/09/pov-tan.pdf_