Google's self-driving cars had a lower crash rate in Mountain View

IIHS researchers found that Google's test cars had 2.19 police-reportable crashes per million miles in Mountain View, against 6.06 for human drivers. The gap was not statistically significant.

  • Not rated society

    The crash rate was lower than for human drivers, but the difference was not statistically significant and employees supervised the cars.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported and intended benefits.
Evidence

foundstudiedreportedclaimed

An independent peer-reviewed study supports the benefit.

Compared with

Police-reported crash rates of human drivers in Mountain View. Measured.

Without the system
6.06 police-reportable crashes per million vehicle miles traveled
With it
2.19 police-reportable crashes per million vehicle miles traveled
The system’s part
Not isolated. The researchers compared Google's test fleet with human drivers in the same city, but employees supervised the cars. The authors could not fully examine how disengagements affected the results. The abstract does not describe how the driving software was built or show that it learns from data.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.

Download data Cite Suggest a correction

Benefit type
Increased Product Safety
System
Google self-driving car (test fleet under employee supervision)
Made by
Google
Used by
Google (test fleet on public roads)
Where
Mountain View, California, United States
Stage
In a pilot
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Operating Vehicles, Mechanized Devices, or Equipment
Last revised
31 October 2026

Sources

  1. Peer-reviewed study, by authors with no tie to the system.

    Teoh ER, Kidd DG. Rage against the machine? Google's self-driving cars versus human drivers. Journal of Safety Research 63:57-60 (2017), doi:10.1016/j.jsr.2017.08.008. Abstract on the bibliography page of the Insurance Institute for Highway Safety, which insurers fund and which has no tie to Google.

    Google cars had a much lower rate of police-reportable crashes per million VMT than human drivers in Mountain View, California, during 2009 to 2015 (2.19 vs. 6.06), but the difference was not statistically significant.

Details

AI system

Workflow

Google tested the cars on public roads in Mountain View under employee supervision. The researchers coded Google's crash narratives by type and severity.

Follow-up

Conditions for replication

The comparison covers crashes severe enough to be reportable to police. The human benchmark comes from police-reported crash rates in the same city.

Entry details

Sector

Road vehicle crash avoidance / automated driving.

Purpose

Google built its self-driving car to drive on public roads, and it tested the cars under employee supervision.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). Google's self-driving cars had a lower crash rate in Mountain View (AIBD-0894). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0894.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions