Incident 22: Waze Navigates Motorists into Wildfires

Description: Waze, a Google-owned directions app, led California drivers into the 2017 Skirball wildfires as they tried to evacuate the area.

Tools

New ReportNew ReportNew ResponseNew ResponseDiscoverDiscover
Alleged: Google developed and deployed an AI system, which harmed Motorists.

Incident Stats

Incident ID
22
Report Count
22
Incident Date
2017-12-06
Editors
Sean McGregor

CSET Taxonomy Classifications

Taxonomy Details

Full Description

Google-owned directions app Waze sent drivers toward areas impacted by the Skirball wildfires in Los Angeles late December 2017. The app looks at current traffic patterns and suggests routes that avoid major congestion. In the case of mass evacuations, as were implemented for these fires, congestion was seen on the evacuation routes leading the app to direct drivers toward the empty roads. These roads were empty because the area ablaze and impassable. Waze/Google engineers typically work with departments of transportation on traffic pattern changes to augment its directions, however in the case of a quickly-developed emergency, the app did not provide safe driving directions.

Short Description

Waze, a Google-owned directions app, led California drivers into the 2017 Skirball wildfires as they tried to evacuate the area.

Severity

Negligible

Harm Type

Harm to physical health/safety

AI System Description

Waze, a directions app meant to decide what route from point A to point B would allow the shortest travel time. Waze also shows potential car accidents, police activity, and other hazards to consider while driving.

System Developer

Google

Sector of Deployment

Transportation and storage

Relevant AI functions

Perception, Cognition

AI Techniques

Interpreting traffic patterns, travel time estimation

AI Applications

Direction suggestions, recommendation engine,

Location

Los Angeles, CA

Named Entities

Waze, Google, Los Angeles Department of Transportation

Technology Purveyor

Waze, Google

Beginning Date

12-2017

Ending Date

12-2017

Near Miss

Near miss

Intent

Accident

Lives Lost

No

Infrastructure Sectors

Transportation

Data Inputs

Traffic patterns, human operator input

Variants

A "variant" is an incident that shares the same causative factors, produces similar harms, and involves the same intelligent systems as a known AI incident. Rather than index variants as entirely separate incidents, we list variations of incidents under the first similar incident submitted to the database. Unlike other submission types to the incident database, variants are not required to have reporting in evidence external to the Incident Database. Learn more from the research paper.