Friday, April 26, 2019

Sluggish or Catastrophic? Five Months of Autonomous Vehicle Non-Progress - I


Since early December, something has been missing.  The weekly and sometimes even daily updates on technical, legal, and organizational driverless car progress have disappeared.  The hottest information-related issue of 2018 (sorry, smartphones), it has been replaced this year by various insights into other applications of artificial intelligence and scattered updates and opinions.  What, though, has been published? 

The most comprehensive article I have seen on this change actually came out near the beginning of it.  Alexis C. Madrigal’s “7 Arguments Against the Autonomous-Vehicle Utopia” (The Atlantic, December 7th), after calling pertinent progress “oversold,” offered a compendium of its largest real and potential obstructions, which was a genuine mixed bag. 

“Bear Case 1:  They Won’t Work Until Cars Are as Smart as Humans” may be why we are lagging behind my most recent forecast (see http://worksnewage.blogspot.com/2018/07/driverless-vehicles-and-driving-jobs.html), with “the sheer number of “edge cases,” i.e. unusual circumstances, they’d have to handle,” and, quoting “legendary roboticist” Rodney Brooks, “perceptual challenges… that are way beyond those that current developers have solved with deep learning networks, and perhaps a lot more automated reasoning than any AI systems have so far [2017] been expected to demonstrate.”  One by one the edge cases can be solved, but if there are indeed vast numbers of them, that will consume some time.  Overall, if the worst issue with driverless technology now is that it isn’t as proficient at dealing with full-contact driving by now as we had anticipated, we’re looking at delays.

Madrigal’s Bear Case 2, “They Won’t Work, Because They’ll Get Hacked,” is a known problem with fixes completed or in progress, and has shown no signs of being uncontrollable.  Bear Case 3, “They Won’t Work as a Transportation Service,” called not only for Uber and Lyft to lose money indefinitely but for them to tire of that and quit, and made the erroneous assumption that “calibrating and maintaining” onboard equipment would be done by individual owners.  The fourth, “They Won’t Work, Because You Can’t Prove They’re Safe,” depends on how people see the 30,000 annual American deaths caused by human drivers, and Bear Case 5, “They’ll Work, But Not Anytime Soon,” ignored that they are functional now, though it seems likely that roll-outs will not be, say, state by state or even city by city, but a square mile or so at a time.  Bear Case 6, “Self-Driving Cars Will Mostly Mean Computer-Assisted Drivers,” confused preliminary progress with the end product.

The most interesting part of Madrigal’s article was “Bear Case 7:  Self-Driving Cars Will Work, But Make Traffic and Emissions Worse.”  As I see it, electric cars have reached a crossroads.  Despite 130 years of technical progress after their first United States production, being beloved by environmentalists (despite over a quarter of their fuel supply coming from coal), being subsidized by the IRS and other parts of the federal government, being facilitated by sufficiently available charging stations at least regionally, and being the recipient of feverish efforts to correct their prohibitive flaw of short battery life, fully electric vehicles still comprise well below 1% of new cars sold, with hybrids sitting at only 2%.  From there, areas recently discontinuing cash grants for their purchase have seen sales drop 75% or more.  Clearly, those working on driverless technology cannot afford to tether it to all-electric platforms.  That means gasoline and diesel engine exhaust not only from people taking driverless trips, but from the vehicles reaching riders.  As for traffic, autonomous cars could end up replacing transit vehicles, and, by providing transportation for more and more people not currently alone in cars due to age or temporary or permanent infirmity, would put additional ones on the road.  All of that means that driverless technology could be seen by most environmental groups as a problem instead of a solution, which could worsen the industry’s second most severe problem, that of legal barriers. 

Three and a half months after that, we got Kara Swisher’s New York Times “Owning a Car Will Soon Be as Quaint as Owning a Horse,” in which the author predicted that she would die before her next auto purchase – meaning, apparently, that she had no plans to move to the country.  Yet the past two months’ other articles were better – see in three weeks. 

No comments:

Post a Comment