AI Simulates Your Daily Life to Test If a Neighborhood Really Works

What makes a neighborhood good to live in? Until now, city planners mostly looked at static measures: How many grocery stores are within a mile? How close is the park? But those numbers don't capture the daily grind of real life. A new study from Haiyan Hao uses artificial intelligence to simulate what a day actually looks like for residents — and the results show that "having stuff nearby" isn't the same as "being able to use it."

The system works like a planning app combined with a smart assistant. It first builds a "knowledge graph" — a digital map linking homes, schools, shops, and road networks. Then it uses large language models (the same kind of AI behind chatbots) to generate realistic schedules for households with different needs: one where a parent must drop off kids before work, another where an elderly person with limited mobility has to do daily errands. The AI checks those schedules for feasibility — like timing and distance — and adjusts them until they work.

In a test neighborhood in Shenzhen, the model exposed a gap between "nominal facility availability" (the number of stores and services listed) and "convenient access" (what residents can actually achieve). Households with care responsibilities — like looking after children or elders — spent much more time coordinating and traveling than the average map would suggest. People with limited mobility faced hidden burdens that standard city statistics simply miss.

The big idea here is that AI can bring "lived experience" into urban planning. Instead of asking residents to fill out surveys, planners could use simulated households to test how a neighborhood might serve different groups. The system is auditable, meaning decisions can be traced back to specific constraints, so it's not a black box. It's early-stage, but it points to a future where cities are designed not just by maps, but by realistic simulations of our messy, busy lives.