Baemin's 'AI Recommended Dispatch' Reduces Rider Accident Probability by 27.8%

Singapore National University Research Team "AI Recommended Dispatch Proves Safety and Income Effects"

Baedal Minjok (Baemin)'s 'AI Recommended Dispatch' has been found to reduce the accident probability of delivery riders by 27.8%, according to a research study. It is analyzed that the AI Recommended Dispatch system, which enhances rider safety and increases income opportunities, needs to be spread across the industry instead of the competitive dispatch method known as 'Jeontu Call'.

On the 16th, according to Woowa Brothers, which is in charge of Baemin's logistics services, Professor Kyung Na-kyung from the Department of Computer Science at the National University of Singapore revealed the safety enhancement effect of the AI Recommended Dispatch system through the study titled "Is the Algorithm a Guardian?: The Impact of AI Recommended Dispatch System on Delivery Rider Safety," conducted at the summer workshop held by the Korean branch of the Association for Information Systems (KrAIS) at Seoul National University.

Baemin Connect Rider

The research team analyzed driving and accident data of riders using the Baemin Connect app for delivery activities through Woowa Brothers. Comparing the accident patterns of riders who used the AI Recommended Dispatch system and those who did not during the same period, the results showed that the accident probability of the rider group using AI Recommended Dispatch was 27.8% lower than that of the group not using it. For example, if 100 riders using the competitive dispatch method have 10 accidents per month, the number decreases to 7.2 accidents in the group using AI Recommended Dispatch.

The reason AI Recommended Dispatch can significantly reduce accident probability is that, compared to the competitive dispatch method commonly used in the delivery industry, it greatly reduces factors that can distract riders, creating an environment where riders can focus on driving. The competitive dispatch method exposes orders to multiple riders simultaneously, and the rider who accepts first wins the dispatch. To secure dispatches, riders inevitably check their smartphones frequently while driving, which prevents them from concentrating on driving and naturally increases the risk of accidents.

Introduced by Baemin in 2020, the AI Recommended Dispatch system proposes optimal dispatches individually based on various information such as the rider’s location and delivery status. In particular, by exposing the dispatch only to the specific rider while deciding whether to accept the dispatch, it fundamentally blocks competition over the same dispatch with other riders, reducing the need for riders to check their smartphones while driving.

The research team explained, "AI Recommended Dispatch has already been proven in previous studies to improve riders' income, and through this study, it was revealed that it not only improves income but also creates an environment where riders can focus on driving, significantly reducing accident probability." Professor Kyung Na-kyung emphasized, "Although many delivery companies still use the Jeontu Call method, the adoption of AI Recommended Dispatch should be expanded across the industry in terms of rider safety and income."

This content was produced with the assistance of AI translation services.