Hong Kong is using an algorithm to schedule and manage nightly engineering work on their subway system.
At midnight, Hong Kong's subway cars go into sidings where a team of engineers do regular maintenance on them. Each week, 10,000 people do 2,600 engineering tasks across the entire subway system. This can include grinding rough nails, replacing tracks or simply checking for damage. Humans carry out the tasks, but the entire process is managed and scheduled by artificial intelligence.
Hong Kong has one of the best subway systems in the entire world, with 99.9 percent on-time accuracy. It is owned by MTR Corporation, who also owns the subway systems in Stockholm, Melbourne, London and Beijing. The company is planning to roll out the AI boss to these other networks as well.
"It will probably be Beijing first," said Andy Chun, the designer of the AI. "Before AI, they would have a planning session with experts from five or six different areas. It was pretty chaotic. Now they just reveal the plan on a huge screen."
The AI works alongside a simulated model of the subway system to find the best way to schedule engineering work. It is able to see where work can be combined and share resources in a way that humans aren't able to. Humans are able to override the system to manually add any unexpected or urgent repairs. The system saves two days a week in scheduling and the engineering team has 30 extra minutes a night to complete their tasks. This saves MTR $800,000 a year.
Chun built the AI with human knowledge that would normally take years of experience to gather.
"We asked the experts what they consider when making a decision, then formulated that into rules – we basically extracted expertise from different areas about engineering works," Chum said.
The algorithm itself works by putting different solutions up against the same issue to see which one is the best fit. The people who carry out the engineering work took awhile to accept the process because they often didn't like knowing why they were performing certain tasks. Workers simply want to know how these decisions are being made.