Waymo’s driverless cars crash way less often than humans: report
https://driving.ca/auto-news/industry/waymo-safer-humans-robotaxi-autonomous-iihs-safetyOpen linkView original on lemmy.todaySyndicated from the fediverse. Read and engage on the original instance.
View original on lemmy.todayhttps://driving.ca/auto-news/industry/waymo-safer-humans-robotaxi-autonomous-iihs-safetyOpen linkView original on lemmy.today
55 replies
I feel like this is a pretty glaring flaw in the study..... They got to redefine what they consider to be a collision and ended up negating nearly 80% of the reports. Something tells me they didn't comb through the data from human reports and do the same.
They'll also kidnap you and deliver you to the police who may or may not murder you depending on their mood.
https://www.latimes.com/business/story/2026-07-07/waymo-reports-teen-riders-for-bad-behavior-delivers-them-to-police
What if you look underage but aren't? What if you take medication but waymo thinks you're doing drugs? There are thousands of scenarios that can be misinterpreted, kidnapping the rider and delivering them to a murder squad is not the solution.
They also park in bike lanes and pedestrian areas so I'm not holding my breath.
That being said, yeah I'm not surprised either. Riding the bus looking at drivers I'll say a solid, very solid 1/4 minimum are doing something with their phones.
Human drivers also park there so at least they're not worse I guess?
The arsehole human parking in the bike lane can be caught and fined.
Allowing Waymo to do it normalises it and makes the issue worse.
Mney where the mouth... uhh,,, would have been is what's needed. Fines for driverless vehicles should be set at, say, four times those for vehicles with drivers, and strict liability for corporate officers for any harm caused to pedestrians, cyclists, other vehicles or drivers.
are you sure waymo is not fined for it?
I don't think they are at the moment, but Waymo seem to expect everyone to just accept it, at least according to some allegations: https://road.cc/news/driverless-taxis-veering-into-cycle-lanes-normal-practice-says-waymo
Enforcement will be difficult like everything else. And I don't like the chances of a local government authority trying to chase Google for fines, seems like it will take a federal government to set the rules as a condition to permitting them to operate.
It's a state matter. Involving the Feds will mean nothing is done.
I feel like Waymos would be easier. You just send it to Google, since they're responsible for operation.
fair
Severe doubt. Getting to decide what a "minor" accident is and exclude it...yeah, uh huh
But block intersections way more than humans. And get stuck in flood zones way more than humans. And pass stopped school busses way more than humans.
And they still have more accidents per mile than actual taxi drivers.
https://guessingheadlights.com/waymo-robotaxis-record-higher-serious-injury-and-fatal-crash-rate-than-nycs-for-hire-fleet/
No, they actually have fewer total accidents per mile than actual taxi drivers. Did you read the link you posted? They are involved in 1.5 times more accidents with severe or fatal injuries per mile, but that is comparing Waymos driving in more open cities at higher speeds to taxis in the very slow speed, high traffic city of NYC where the primary use case is different. Also, the only fatal accidents (2 of them) have been the fault of other cars, not Waymos, so it's kinda unfair to include those.
“The researchers compared Waymo’s nationwide crash data with New York City’s for-hire fleet and found the robotaxis had a serious injury or fatal crash rate about 1.5 times higher per mile driven.”
Yes, thank you for quoting exactly what I said??
You are quoting the serious or fatal injury stats. The overall injury accident rate, though, as quoted from the article: "When all injury-causing crashes were considered, Waymo performed better. Its robotaxis were involved in 67% fewer injury crashes per million miles than New York City's for-hire vehicles."
Fatalities and serious injuries sound like a far more important statistic to optimize for in terms of public safety.
Looking at the raw data from NYC, in 2025 there were 15 crashes with fatalities and only 3 with serious injuries. To follow your train of thought, I would think the survivability of a serious crash is important to optimize for, and Waymo has a better ratio for that (4 to 2 versus 3 to 15).
Agreed. I'm just correcting misinformation and giving context.
They would need to have more of them first for the data to actually be accurate at all. From what I could find, there have been no Waymo caused fatalities - two deaths caused by other drivers though - and four serious injuries, for roughly 220 million miles driven.
While you can then state that the rate of fatalities is 1 per 110 million miles, it's not really useful in any way.
Wow you're trying really hard to shill for these AI cars. Imagine downplaying a 50% increase in death rate
Funny how you ignored this part too:
I'm not shilling for AI cars, I'm correcting idiots who either don't know how to read or who intentionally give incorrect information.
You, for instance, just said "imagine downplaying a 50% increase in death rate." That is false. It's a 50% increase in crashes involving death OR a serious injury. Is your reading comprehension bad and you didn't notice that it was for both types of crashes, or are you intentionally giving bad information to try to make a point? Additionally, there have only been 2 crashes involving Waymos where people have died. BOTH of those crashes were the fault of the human driver in the other car. One of them was the Waymo sitting stopped at red light and it got rammed from behind by a drunk driver in an SUV, which pushed the Waymo into the intersection where another car hit it. There have also only been 4 Waymo crashes with serious injuries in something like 220 million miles. I can't find whether those 4 crashes were the Waymo's fault. You seem caught up on this 50% increase number, so if we compare it to taxi drivers, that means they had 4 crashes per 220 million miles that had either a death or serious injury, compared to Waymo's 6. But the problem is the statistics aren't comparable. There were 15 fatalities from taxis in NYC in 2025 alone, versus 2 fatalities in Waymos total. There are just far far far more millions of miles driven to pull data from. It's also comparing apples to oranges with where each group is driving. You are probably going to argue that, but looking at the NYC data, 11 of the 15 fatalities were involving Black Cars that normally take people to the airport and will be driving faster versus medallion cabs that are inside the city at much lower speeds. Waymo operates in faster driving cities with less traffic than medallion cabs, so they'll be expected to have more serious injury crashes. Also interesting: in 2025 there were 15 fatalities with taxis in NYC, but there were only 3 crashes with serious injuries. You are more likely to die in a crash in a taxi than get a serious injury, but in a Waymo you are more likely to survive a serious crash.
As far as me ignoring minor incidents that would not have been reported to police, why would I bring that up? There is no information to compare. Waymo might have that data, but the taxi drivers in NYC and human drivers all around the world do not report and so we will not have the data to compare to. Why are you ignoring minor incident data from taxi drivers? Are you shilling for Yellow Cab?
so NYC is safer than other places where waymos drive
Yeah, Filipinos are decent drivers.
Waymo admits that its autopilot is often just guys from the Philippines
That explains why they park illegally like all the time.
When are we gonna see waymo motorbikes?
That isn't really what that article says. It says that in edge cases, remote workers (mostly in the Philippines) give additional contextual inputs to the autopilot. There are some outstanding issues with waymos, especially around getting stuck idling or infinite loops at construction sites, but really just being remotely operated is not one of them.
That's cool, can we get public transport please?
Without knowing the methodology the IIHS used to arrive at its claim, I'll remain unconvinced.
so they don't crash way mo?
I'll see myself out
And if they do they crash way mo slo
hot take but i'm actually excited for these to come to seattle
can't get sexually harassed by a car with no driver. this is a recurring problem when i take ubers
A waymo recently locked its teenage passengers in the car because it guessed they might be underage drinking, called the cops, and delivered them to the police station
Source on this? that is insane.
https://www.latimes.com/business/story/2026-07-07/waymo-reports-teen-riders-for-bad-behavior-delivers-them-to-police
Oh yes you can. Shareholders have decided an upskirt cam they sell access to online is a perfectly acceptable revenue stream that will double as a Republican feature to ensure the trans folks aren’t in their cars.
We think you’ll love it.
Sign in with the Waymo app and link it to your health data to track your cycles and detect early infections with visual recognition software from out upskirt cam.
$14.99/mo subscription
$29 a month without ads (we only consider it an ad if advertisers didn’t pay us enough to force it into this tier anyway)
How many accidents do they cause?
How often do they interfere with traffic our the standard operation of transport by just sitting there.
Waymos on YouTube getting stuck in places cars shouldn't be is about as good an example that exists that these should all be ADAS features in cars with the human incharge, and that AI should also be used as a tool and not to replace humans.
The comparison isn't fair if the human driver doesn't have similar ADAS features. But nice try Waymo.
Nope, not buying it. Ai makes mistakes way more than regular people do
You do realize those cars are not using LLMs to drive, right?
whatever type of AI they use, it definitely makes mistakes all the time
Yes, but you didn't say "make mistakes". You said AI makes way more mistakes than regular people. And that's simply not true:
Sure lol. It's not true when you use cherry picked stats, for cars that only drive in areas literally curated for them.
Seriously? People as a group suck at driving. The bar is super low. You are just going to form a random opinion on this and ignore data?
I dunno, regular people can barely handle walking sometimes. Driving is especially hard.
X doubt
not hard to believe at all
waymos do silly things, but people do dangerous things more often like using their phones, get distracted, driving angry/tired, road rages at rush hour, speeding, etc
Lmao the article literally admits the data is cherry picked and only applied on "a limited scale"
It's a limited scale because waymo only operates in bounded areas, not because they cherry picked the data
This is a pretty good video that goes into the intricacies of it
https://youtu.be/2DOd4RLNeT4
Contradicts the previous articles stating the opposite. (Easy to find, I don’t provide search services. )
if waymos were as bad as the average human driver they'd already be out of business
Would they? lol. I’m not sure that’s how 2026 works.
fair, they'd have pivoted to data centers or something