Improving public transport disruption communication
Passengers receive late or inconsistent disruption messages across apps, signs, and station staff.
Problem
When incidents occur, operations teams manually rewrite messages for channels with different length and language needs.
Current workflow
Control center notes an incident, drafts text, pushes to some channels, and relies on staff for the rest.
Consequences
Crowding, missed connections, and eroded trust in transit agencies.
How AI might help
Assistants can draft consistent multi-channel messages from incident codes and suggest alternative routes.
Limits & risks
Need accurate incident state; hallucination risk for specific station names.
Incorrect diversion advice; alarming tone that increases panic.
Alternatives today
Template libraries and vendor GTFS-realtime feeds.
Templates are incomplete; feeds lag human narrative detail passengers need.
Evidence
- UITP passenger information practices
Documents real-time information as a core service quality driver.
Comments
No comments yet.