Singapore AI Predicts How People Spend Government Handouts

Government officials might soon know exactly how citizens will spend their money before handing out a single dollar, thanks to a breakthrough AI tool from Singapore.

Four researchers at Nanyang Technological University developed an artificial intelligence system that predicts whether people will save or spend government cash handouts and what they'll buy with that money. The tool analyzes real spending patterns instead of relying on surveys where people often say one thing but do another.

The project stood out among 400 submissions worldwide and won $100,000 in funding from OpenAI. Professor Hyeokkoo Eric Kwon leads the team that will spend the next six months perfecting the system.

The AI learned from transaction data of over one million users of a South Korean budgeting app, tracking their spending between 2023 and 2025. All personal information was removed, and the team built strict safeguards requiring at least three to ten users to share identical characteristics before any profile appears in the system.

Here's what makes this different from traditional policy planning. When governments consider giving everyone $100 in vouchers, they usually rely on interviews and surveys to guess the impact. But people often behave differently when actual money hits their accounts compared to how they respond in hypothetical questions.

The AI runs simulations showing where citizens might spend the money and how much economic activity different policy designs could generate. The team will test their predictions against real spending data from two rounds of South Korean government handouts distributed in July and September 2025.

The Ripple Effect

This technology could transform how governments worldwide design economic support programs. Instead of guessing which policies help communities most, officials can test different approaches virtually and choose the ones that create the biggest positive impact.

The researchers plan to make the tool open source by February 2027 so any country can adapt it for their population. The system will explain its reasoning for each prediction, helping policymakers understand not just what citizens might do but why they might make those choices.

Better policy design means more efficient use of taxpayer money and programs that actually meet people's needs. A government could discover that cash works better than vouchers, or that timing matters more than amount, before spending millions on a program that misses the mark.

The gap between good intentions and good outcomes in government policy might finally have a bridge.