AI Innovations in Ride-Hailing: Didi’s Smart Brain Applications

On August 15, Didi’s CTO Qie Xiaohu delivered a keynote titled “AI Technology Innovation in Mobility: Opportunities and Challenges” at the IJCAI conference in Macau, sharing how Didi integrates AI across its ride‑hailing platform.

The presentation highlighted Didi’s “Smart Brain”, which embeds AI in every stage of a user’s journey—from destination search, ETA estimation, and pick‑up point recommendation to order allocation, driver dispatch, and route planning.

Didi has evolved its dispatch system from simple proximity‑based matching to a globally optimal intelligent dispatch. This not only maximizes real‑time efficiency but also incorporates future demand forecasts, resulting in a 2% increase in drivers’ long‑term earnings.

AI techniques have also reduced Didi’s ETA error rate to 11% by leveraging massive historical trip data, real‑time traffic conditions, and weather information through deep learning, enabling second‑level decision making and multi‑strategy route planning.

The company introduced a “route selection” feature that presents passengers with up to three route options, including distance, estimated arrival time, and traffic signal counts, while automatically syncing the chosen route to the driver’s navigation system.

Beyond efficiency, Didi emphasizes adding a “human touch” to AI, focusing on user psychology and experience. The AI‑driven solutions cover pick‑up point optimization, intelligent driver scheduling, idle‑time compensation, route protection for short‑haul airport trips, and more.

Didi’s ecosystem now offers the Smart Brain and the “Qun Yan” intelligent travel sharing platform to city traffic managers, smart‑transport enterprises, automotive partners, developers, researchers, and NGOs, providing customized open services.

Additionally, Didi continues to mine large‑scale mobility data with machine‑learning methods to identify safety risks and improve overall travel safety.