Across Asia, artificial intelligence (AI) technology is being deployed in the social sector faster than it can be absorbed. A tool that may perform well in a demonstration might still never reach the people it was built for, unable to cross from the pilot stage into a sustainable initiative.
This research from the Centre for Asian Philanthropy and Society (CAPS) draws on a case database of 220 initiatives across Asia Pacific and interviews with 42 organizations in 10 economies to find that the sector keeps buying “cars” without building “roads”. Most private capital being deployed for AI for Good is going to the visible tools and applications that are easy to fund and demonstrate, while the human capability, shared infrastructure, delivery channels and governance that turn a pilot project into lasting public value remain underfunded.
Four key findings stand out:
- Readiness decides success more than the model. The primary constraint in AI adoption is human and organizational readiness, not technical capability.
- Success carries a penalty. Many AI-driven services incur costs every time they are used, so expenses continue after adoption and can rise with scale – a poor fit for funding built on fixed grants and one-time deliverables.
- Infrastructure is primarily a coordination problem. The shared datasets, platforms and delivery channels that make AI scalable benefit entire ecosystems but rarely make sense for any single organization to build alone.
- In practice, ethics is decided before the build. Whether AI serves people fairly is determined at the design stage, long before governance can intervene.
The report identifies priorities for private social investors to sustain impact: build human infrastructure, co-create with beneficiaries, develop shared infrastructure, restructure the economics of funding and require ethics by design. It offers practical guidance for funders seeking to support not only AI applications and tools, but the operating environment that determines whether such initiatives succeed.
The AI for Good Navigator
The report is accompanied by the AI for Good Navigator. This interactive site maps 220 initiatives from across Asia collated for this research, allowing readers to explore the report’s findings, search cases, follow different use cases for deployment and learn from others’ experiences.
Across Asia, artificial intelligence (AI) technology is being deployed in the social sector faster than it can be absorbed. A tool that may perform well in a demonstration might still never reach the people it was built for, unable to cross from the pilot stage into a sustainable initiative.
This research from the Centre for Asian Philanthropy and Society (CAPS) draws on a case database of 220 initiatives across Asia Pacific and interviews with 42 organizations in 10 economies to find that the sector keeps buying “cars” without building “roads”. Most private capital being deployed for AI for Good is going to the visible tools and applications that are easy to fund and demonstrate, while the human capability, shared infrastructure, delivery channels and governance that turn a pilot project into lasting public value remain underfunded.
Four key findings stand out:
- Readiness decides success more than the model. The primary constraint in AI adoption is human and organizational readiness, not technical capability.
- Success carries a penalty. Many AI-driven services incur costs every time they are used, so expenses continue after adoption and can rise with scale – a poor fit for funding built on fixed grants and one-time deliverables.
- Infrastructure is primarily a coordination problem. The shared datasets, platforms and delivery channels that make AI scalable benefit entire ecosystems but rarely make sense for any single organization to build alone.
- In practice, ethics is decided before the build. Whether AI serves people fairly is determined at the design stage, long before governance can intervene.
The report identifies priorities for private social investors to sustain impact: build human infrastructure, co-create with beneficiaries, develop shared infrastructure, restructure the economics of funding and require ethics by design. It offers practical guidance for funders seeking to support not only AI applications and tools, but the operating environment that determines whether such initiatives succeed.
The AI for Good Navigator
The report is accompanied by the AI for Good Navigator. This interactive site maps 220 initiatives from across Asia collated for this research, allowing readers to explore the report’s findings, search cases, follow different use cases for deployment and learn from others’ experiences.