James Joyner has a good roundup on Google’s work in creating an autotaxi, a subject I have posted on before. It’s not exactly the same because my scenario involve “antcars” which were all entirely driven by automation while these have manual override, but you don’t get one without getting the other first.
A few observations:
- One of the links talks about who the big losers would be and singles out truckers. I would add another huge, huge loser: Public transportation. Other than being unable to afford a car (not applicable to most of the population) or traffic being so bad that it’s not worth it (pretty rare), one of the huge benefits of public transportation is that you don’t have to pay attention to the road. With autotaxis, you would get that benefit while also being able to go wherever you want and you would get to avoid the kinds of people that take public transportation. When I had jury duty, I took the commuter bus from Mayne to downtown and most of the riders were white-collar individuals. Most. However, throw in autotaxis and most of those people would probably prefer that to driving to the bus station, waiting for a bus, and then walking from their stop to their final destination (or worse, having to change buses).
- In my antcar post, Peter commented that these things would be a lawsuit waiting to happen. I think in the autotaxi phase this is true (by the time they become antcars, this has presumably been taken care of). It stands to reason that even in the event that the technology is invented here, we may be the last country to have widespread adoption. The makers would charge huge prices in preparation for the eventual lawsuit. As one of the cited links mentions, we are always much more anxious to blame technology when there is any ambiguity and the Toyota carnival is the perfect example of this.
- I stand by my scenario for antcars insofar as I believe that if these things ever achieve widespread adoption, you can very much expect the government to demand that records be kept for later subpoenas. They will probably justify it on safety grounds, arguing that it’s important to know what your car was doing in the event of an accident. Or something. They’ll almost certainly find some reason why it is technically necessary for some benign reason. Sort of like how Microsoft build IE into the fabric of Windows so that they could later claim that it can’t come not-preinstalled. The government will want it, but the makers probably will too in order to fend off future lawsuits (someone blaming the auto-drive for manual error). The only people against it will be the consumer, and what’s that good for?
- As you might have been able to tell from my antcar scenario, I think that Joyner’s question of how long human drivers will be allowed on the road once the safety of the computers surpasses that of human drivers is spot on. Once that threshold is reached, I really don’t know how long it will be until at least some people start agitating for it. Probably the generation that starts immediately with the autotaxis will be the generation that won’t see mandatory antcars as no big deal. Until then it will be the subject of contentious debate.
- Could the antcar ever comes to pass or autotaxis become completely automated and so ubiquitous that most people don’t need driver’s licenses? Driver’s Licenses are currently one of the primary incentives to get IDed by the government. What then?
Category: Road
About the Author
6 Responses to Antcars, Redux
Leave a Reply
please enter your email address on this page.
Antcars on purpose-made lanes, doing tasks like delivery of goods and commuters, is one thing.
Antcars supplanting all human drivers is far less likely – like a lot of other technological scenarios, by the time the applicable tech reaches that point (or more precicely, when the AI required to make it work reaches that point) we will have a completly different world around us.
Barring human-level AI, antcars probably won’t replace humans completely for the basic reason that a lot of people LIKE driving and won’t give it up completely jst because they are asked t.
They might very well let antcars take over the kind of driving that people need to be paid to do, like mail delivery for instance.
I would add another huge, huge loser: Public transportation.
The problem is that it still doesn’t solve the other inherent issues with automobiles: the need for parking lots and garages, traffic, and the fuel to power these vehicles. The easiest way to create public transport demand is to restrict parking and not build large freeways to induce traffic while increasing the cost of ownership & operation. Even if we don’t do that, ulimately, I’d question if it’s possible to expand highways even more to move slightly more traffic when the capacity of a light rail line is equivalent to that of IIRC, several lanes of traffic while using far less space.
Of course, the other problem is that when these cars fail in the hands of people who barely drive, we’re going to be stuck with people with poor driving skills and little experience attempting to prevent accidents…
I think that Joyner’s question of how long human drivers will be allowed on the road once the safety of the computers surpasses that of human drivers is spot on
You’ll pry my freedom to drive without your pathetic robots from my cold, dead hands. I want to live in a country for drivers, not a country for cars…
Barring human-level AI, antcars probably won’t replace humans completely for the basic reason that a lot of people LIKE driving and won’t give it up completely jst because they are asked t.
I don’t think human-level AI is remotely required. What they lack in intellectual nuance, they make up for with the ability to consistently pay attention to a wider variety of things. They won’t get distracted and they won’t get drunk. They’ll never be better than the best drivers, but it won’t take them long at all to be significantly better than most drivers.
I agree about people liking driving. The conversation on my Antcars post got sidetracked primarily to questions of privacy, but the desire to drive is significant. However, what happens when it can be demonstrated that automobile accidents will decrease by some 90% if we turn over our keys? I think the joy of driving will become a tougher sell. This part is pretty far down the line, though. But I think that chances are better than not that the day will arrive.
The problem is that it still doesn’t solve the other inherent issues with automobiles: the need for parking lots and garages, traffic, and the fuel to power these vehicles.
If one believes that we are rapidly approaching peak oil and/or that public transportation will replace automobiles as the primary mode of transportation out of necessity, antcars won’t change anything. I do not believe either of these things to be true.
Even if we don’t do that, ulimately, I’d question if it’s possible to expand highways even more to move slightly more traffic when the capacity of a light rail line is equivalent to that of IIRC, several lanes of traffic while using far less space.
The issue here is that rail takes you through a pretty limited number of paths. Cars can go anywhere and have an unlimited number of stops along the way.
Of course, the other problem is that when these cars fail in the hands of people who barely drive, we’re going to be stuck with people with poor driving skills and little experience attempting to prevent accidents…
Driving is a pretty intuitive thing. Most people have a pretty basic idea of how to do it before they ever get behind a wheel. The learning process is geared towards the complex rules of the road. You don’t need to know that as much in order to avoid accidents if your car fails.
The bigger concern I have is that cars will fail without people realizing that they’re failing. They’re playing on the phone-gizmo or whatever while their car’s breaking or buffer mechanism is quietly locked up. I think these things will work themselves out, though.
One of the larger problems with antcars is, how do you get them to manage variable road conditions? Which is to say – ice, black ice, snow, rain-slick, post-rain, dry, dusty, dirt-covered, ill-repaired, cobblestone, and just plain dirt roads?
In there is enough variation, even without requiring “human-level” emergency reaction AI, to handle requiring a myriad of sensors and controls and redundant hardware design and complicated enough AI to give modern processors – much less the ones that would be required to operate in the extreme temperature ranges of an engine – to give computer programmers and safety experts nightmares for a couple of decades if not millenia.
Web, I think we’re looking at three phases: manual-autotaxis (you can take over the car when need-be), individual antcars (cars without drivers), and mandatory antcars (no drivers allowed).
I think the things you’re concerned about get addressed in #1. People use the auto function the same way that they currently use cruise-control. Turn it on when you’re comfortable and leave it off the rest of the time. The push among automakers will be to make it so that you can have it off a greater and greater majority of the time. They’ll start figuring out what to do about ice and snow at that point. First a button or maybe a dial and it goes into a increasingly “cautious” modes when you turn on windshield wipers. Then, more and more, it’ll already know what the roads are like.
Phase 2 comes around when they’re good enough that you don’t need an actual driver. Pretty limited at first to taxi and courier operations, but eventually it will be affordable to the rest of us to get the cars that need no input from us.
By the time we reach Phase 3, the cars will know the roads a lot better than we will because they will have up-to-date reports from some central service keeping track of it. I don’t know that we will ever get to this phase. If so, what will likely happen is that there will be antcar-zones where manual is not allowed and these zones will expand ever-outward. If this happens in the US, it probably will not be in our lifetime.