SoylentNews is people
SoylentNews
from the darn-it-someone-must-take-the-blame dept.
Uber's self driving car program in AZ isn't out of the woods yet. The Phoenix New Times https://www.phoenixnewtimes.com/news/claim-ducey-state-blamed-uber-self-driving-death-unsafe-policy-11205678 reported last month that lawyers representing family of Uber victim Herzberg have sued the state of AZ for $10M, fingering Governor Ducey for failing to protect the people of his state.
After quoting legal precedent about the state's responsibility to keep roadways "reasonably safe" for travelers, the claim says the state has "failed to make roadways safe, allowing autonomous vehicles to operate on public roadways in an unsafe manner."
The state's oversight of autonomous vehicles was negligent, it states, adding that Ducey's 2015 executive order facilitating the testing of self-driving vehicles was created "negligently and without sufficient investigation into the safety of Uber's autonomous vehicles. Any oversight provided by a committed, ADOT, or DPS, was wholly insufficient, and placed an unreasonably high risk of harm to the citizens of Arizona."
The claim goes on to quote Ducey's 2016 invitation to Uber, in which the governor quipped that "California put the brakes on innovation and change," but he wouldn't. "This rush to be first in the 'tech boom' era made Arizona's roadways unreasonably dangerous," the claim states.
New Times made a similar argument that Ducey was at least partially responsible for Herzberg's death in the April 12 cover story, "Ducey's Drive-By: How Arizona Governor Helped Cause Uber's Fatal Self-Driving Car Crash."
Last time Gov. Ducey appeared here on SN was back on December 03 2016, https://soylentnews.org/article.pl?sid=16/12/02/2341241 He announced that Lucid would start making cars in a new plant in AZ...in 2018. Looks like he missed that prediction --
Officials from electric vehicle startup Lucid Motors, which broke cover from stealth mode in October [cnet.com], made a joint appearance today with Arizona Governor Doug Ducey and Sonora, Mexico's Governor Claudia Pavlovich Arellano, to announce a manufacturing plant in Casa Grande, Arizona. The plant will begin production of Lucid's first car, an electric luxury sedan, in 2018, with parts being supplied from across the border in Sonora, Mexico.
Governor Ducey said the new plant will create 2,000 jobs by 2022 [cnet.com], and that Lucid Motors has promised to prioritize hiring among Arizona veterans.
Lucid Motors has shown a very sophisticated operation for its entry as a new automaker, with its Chief Technology Officer, Peter Rawlinson, an alumni of Tesla and Lotus, and Vice President of Design Derek Jenkins having spent time at Mazda and Volkswagen. The as-yet unnamed first model will compete with the Tesla Model S as a luxury sedan, and should boast over 300 miles of range. Lucid has also designed connected features and self-driving capability into this car.
(Score: 0) by Anonymous Coward on Monday March 18 2019, @05:42AM (8 children)
Good drivers are what, 100x safer than the average? Let's be generous here and assume that.
How many times safer are average drivers compared to people who've never driven (without instruction)? Easily 1000x or so in terms of accidents/unit distance.
How many times safer are people who've never driven compared to toddlers? Easily 1,000,000x or so.
The difference between an untrained AI and a shitty driving AI is vastly larger than the difference between a shitty driving AI and a better-than-unmodified-humans-could-ever-be AI.
The scale isn't |toddler----------------untrained----------------average----------------good----------------superhuman|, it's probably far closer to |toddler------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------untrained--------------------------------------------------------------------------------average---------good-------superhuman|
Once AI is as good as a barely competent driver, we're nearly finished making it utterly superhuman in terms of total effort. Don't let the slow years ahead overshadow the literal decades which preceded them.
Lameness filter encountered. Your comment violated the "postercomment" compression filter.
Try less whitespace and/or less repetition.
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Morbi dapibus vel velit eget lacinia. Integer sed posuere lorem. Ut nec lobortis metus. Vestibulum vitae vulputate ipsum. Quisque suscipit ante velit, sed volutpat arcu suscipit dictum. Phasellus vitae sem vel nunc faucibus feugiat rutrum vitae erat. Go fuck yerrself ya crappy heuristic. Donec varius luctus arcu, id commodo urna dictum varius. Morbi id ipsum a massa varius laoreet eget at ligula. Vestibulum nisi turpis, aliquet luctus consequat id, faucibus nec justo. Duis lobortis, nulla laoreet accumsan tristique, mi lectus aliquet mi, in laoreet dui nibh vestibulum sem. Nunc varius urna id metus tristique, nec auctor orci iaculis.
(Score: 0) by Anonymous Coward on Monday March 18 2019, @05:54AM
I don't know much at all about AI research, I could be totally wrong, this post is pure speculation, and I didn't notice the tone I was writing it with until I posted. Please imagine I sprinkled some "I expect"s and "it would seem"s in it and didn't write as if I was stating facts.
(Score: 0) by Anonymous Coward on Monday March 18 2019, @07:39AM
Your PS being noted, you also forgot to account for the law of diminishing returns.
(Score: 1) by khallow on Monday March 18 2019, @03:17PM (5 children)
(Score: 0) by Anonymous Coward on Tuesday March 19 2019, @01:58AM (4 children)
Cars could get us around far faster if their limitations were the max accelerations our bodies could handle rather than the max speed we can be safe at.
(Score: 1) by khallow on Tuesday March 19 2019, @03:41AM (3 children)
Indeed. But that's my point. You can't get past those limitations so easily. The perfect AI couldn't brake much faster than a trained human would, for example. So if it's tailgating another perfect AI which has a tire blowout, it's still a multi-car accident.
(Score: 0) by Anonymous Coward on Tuesday March 19 2019, @06:12AM (2 children)
A superhuman AI would provide economic incentive for better brakes/handling/acceleration that just isn't valuable to consumers right now except as bragging rights.
Also, when one's reaction times are measured in microseconds and one can run crude physics sims in milliseconds to check if a maneuver works, a whole lot of dodging around accidents becomes possible that humans couldn't even attempt without losing control and making it worse.
(Score: 1) by khallow on Tuesday March 19 2019, @12:13PM (1 child)
I strongly doubt it. Particularly, for vehicles that are just rented out which is the common self-driving scenario.
And physical lag that often can be seconds long.
Keep in mind the growing sensitivity to liability in developed world societies, particularly when there are deep pockets around. If a high energy accident happens in a packed highway, and it becomes a matter of who gets to experience harm (since in the seconds of evolution of the accident not everyone can fit in safe phase space zones around the accident), every step of that above process becomes something that can be questioned in court. Thus, there will be considerable slack in traffic flow, if only to reduce one's eventual court costs.
(Score: 0) by Anonymous Coward on Tuesday March 19 2019, @05:52PM
It's an interesting ethical question too.