Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

sfchronicle.com

Post reply on HN
Print view View original

111–120 of 1001 posts

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#111

Post by __loam »

paxys wrote: post #92
parl_match wrote: post #3

I'm wary about using public roads to test these, but I think the way the data is presented is misleading. I'm not sure how it's misleading, but separating "incidents" into categories (safety, traffic, accident, etc) might be a good start. For example, I could start coning cruise cars, and cause these numbers to skyrocket. While that's an inconvenience to other drivers, it's not a safety issue at all. By the way, as a…

> For example, I could start coning cruise cars, and cause these numbers to skyrocket And why would that not be a valid data point? Cars driving on public roads need to successfully handle all regular day-to-day situations, not just the happy path. It seems like almost every day I run into some Cruise or Waymo (mostly Cruise) vehicle stuck in the middle of the street blocking traffic, and that should not be excused a…

It's a lot higher than 0.1 people. 40 people died in SF last year due to traffic accidents. I've yet to hear about a self driving car killing someone despite how many there are in the city.

I also disagree that coning a self-driving car's sensor is a "regular day to day situation". Throwing paint on someone's windshield or slashing their tires is a crime.

E: making stat accurate

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#112

Post by brigade »

paxys wrote: post #104
veec_cas_tant wrote: post #102

Earlier quoted context omitted.

> And why would that not be a valid data point? Cars driving on public roads need to successfully handle all regular day-to-day situations, not just the happy path Are you considering someone putting a traffic cone on a vehicle a day-to-day situation?

What would you do if you were stopped at an intersection and someone put a traffic cone in front of you? Sit there for the rest of your life blocking everyone behind you, or move it/drive around it? Same for literally every example that deviates from the happy path, no matter how minor. Construction work, something fallen on the road, a delivery van stopped on the side, temporary street closures. These cars are nowhe…

They put the cone on the hood, it's not an obstacle on the road. The car could try to shake it off, maybe, but is shaking it into the middle of the street and creating an obstacle really the right move? Should self-driving cars be required to have a robotic arm capable of moving a traffic cone from a car surface to the sidewalk to handle this specific situation?

Also the anti-car activists can easily move on to a wide variety of other methods to obstruct sensors.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#113

Post by talldatethrow »

DragonStrength wrote: post #14

I left SF a year ago but had Cruise and Waymo cars in my neighborhood constantly. They were a total nuisance. They'd jerk towards pedestrians in crosswalks leading to this uncomfortable standoff where no one in the neighborhood would ever step in front of one. I'd wager any stats here are way lower than real incidences as I was nearly hit multiple times and never got to the "report it" stage as there are so many barr…

New sport.. anyone in need of cash throws caution to the wind and temps ai cars to hit them for the payout.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#115

Post by dekhn »

"Skyrocket" is a word people use when they don't have the data to support their claim, but want to make it sound as extreme as possible.

When I hear "skyrocket" I think "several orders of magnitude in a very short time". But the source of the quote with skyrocket, basically admits that they didn't have any quantitative evidence to support a massive change in incidents.

Remember: reality is banal. Things are far better explained by sampling error, bias, and base rates, than they are by sudden, dramatic shifts caused by a single factor.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#116

Post by Animats »

parl_match wrote: post #3

I'm wary about using public roads to test these, but I think the way the data is presented is misleading. I'm not sure how it's misleading, but separating "incidents" into categories (safety, traffic, accident, etc) might be a good start. For example, I could start coning cruise cars, and cause these numbers to skyrocket. While that's an inconvenience to other drivers, it's not a safety issue at all. By the way, as a…

> For example, I could start coning cruise cars, and cause these numbers to skyrocket.

Google could get tough about that. If you have Google Play Services on your phone, they know who you are. "Your Google account has been terminated for violation of Google's terms of service."

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#117

Post by abeppu »

reso wrote: post #63
valine wrote: post #47

Earlier quoted context omitted.

Why does it require AGI? That's a bold, unsubstantiated claim. Let me make my own bold claim: Simple common sense reasoning is very close to being solved with our latest LLMs. You would need AGI if you wanted to replace the team developing your self-driving car. A decent LLM, with vision input that's been fine-tuned on driving scenarios, will be more than sufficient for Level 5 driving.

You need AGI for good self-driving because driving requires predicting the actions of other human drivers at an extremely high level. This is second-nature for humans so we barely notice that we are doing it, but it is extraordinarily difficult for non-humans.

I think the reality is somewhere in the middle. You need to be able to accurately predict behavior of humans _following some conventions_, and to be wary of the behavior of humans when they violate those conventions.

An example I saw:

- At the start of a construction area, a guy wearing a hi-viz vest holds a stop sign. A self-driving car stops at the sign.

- The guy _lowers_ the sign a bit while looking over his shoulder down the street towards others on his crew.

At this point a _human_ guesses that the sign is lowered only b/c the guy has seen that the car stopped, and expects the car to stay stopped until some further signal (e.g. a waving gesture, or flipping the sign to show the "slow" side). The human driver understands that stop sign guy is looking to coordinate with someone else nearby. There's a "script" for this kind of interaction.

... but the self-driving car starts moving as soon as the road crew guy lowers the sign. In this case nothing seriously bad happened. But it was not following The Conventions.

This doesn't take full general intelligence perhaps -- but it takes some greater reasoning about what people are doing than the cars seem to have currently, and so sometimes they drive into a zone that the fire department is actively using to fight a fire, and get in the way.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#118

Post by tlb »

Eisenstein wrote: post #94
standardUser wrote: post #22

Earlier quoted context omitted.

I know this is a tired argument, but it's not wrong: There is equal or greater cause for concern with cars driven by people, we're just used to those horrors and reluctant to trade them for a different set of horrors, even if they are conclusively less horrible. I lived carless in downtown SF for 10 years, usually with a walking commute, and it was not pleasant. Packed streets full of angry, stressed drivers honking…

Depending on driverless taxis instead of having emotional people who need parking all the time seems to be a win, as long as they are safe and someone is held accountable when they mess up . I am very unwilling to change 'people are flawed but (mostly) held accountable' to 'a corporation owned robot car ran someone over because of a software glitch and it is nobody's fault'. The problems with 'oops we got hacked or l…

Suppose it's the near future and driverless cars are causing 5x fewer injuries per mile than human drivers. (5x fewer != 0, so some injuries and deaths still occur)

What accountability would you like to see in that situation?

Or a table of

  | relative death rate | accountability |
would cover the widest range of future scenarios.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#119

Post by addisonl »

yazaddaruvala wrote: post #56
laweijfmvo wrote: post #24

What seems odd to me is that hardly anyone in SF seems to want them there (citizens, officials) yet it keeps getting pushed through?

I'm so excited for them to come to Seattle! There are many people that want them! There are even more than want 90% cheaper Ubers.

Wouldn’t get your hopes up on it being 90% cheaper—typically Waymo cost the same or slightly more than Uber for me. Obviously the economics change as time goes on but I doubt they will make it that much cheaper.

Re: S.F. says incidents by Cruise, Waymo driverless taxis are ‘skyrocketing.’

#120

Post by dekhn »

reso wrote: post #63
valine wrote: post #47

Earlier quoted context omitted.

Why does it require AGI? That's a bold, unsubstantiated claim. Let me make my own bold claim: Simple common sense reasoning is very close to being solved with our latest LLMs. You would need AGI if you wanted to replace the team developing your self-driving car. A decent LLM, with vision input that's been fine-tuned on driving scenarios, will be more than sufficient for Level 5 driving.

You need AGI for good self-driving because driving requires predicting the actions of other human drivers at an extremely high level. This is second-nature for humans so we barely notice that we are doing it, but it is extraordinarily difficult for non-humans.

I don't think anybody has demonstrated convincingly that a self-driving car would specifically need AGI to acheive what really matters: statistically better results on a wide range of metrics. I don't expect SDCs to solve trolley problems (or human drivers to solve them either) or deal with truly exceptional situations. To me that's just setting up an unnecessarily high bar.
Post reply on HN