Driverless Cars: Not as Autonomous As We Thought

  

As part of an investigation by U. S. Sen. Edward Markey (D-Mass.), autonomous-vehicle operators Tesla, Waymo, and others have made public some details about how their supposedly driverless cars can be remotely assisted or controlled under certain circumstances.  In a report published on Apr. 8, Andrea Guzman of the Austin American-Statesman described how the companies reluctantly shared some but not all the information Sen. Markey requested.  Specifically, they refused to say how often such interventions take place.

 

It was a surprise to me, though it probably shouldn't have been, that autonomous-vehicle firms have staffs of remote operators ready to take control of a car or truck that gets into trouble.  It's a given that every such vehicle is in continuous wireless communication with "headquarters," whatever that amounts to.  And when the onboard systems determine that the situation the vehicle is facing is beyond its capabilities, it's reasonable that it sends an alert to headquarters to flag a human being to take a look and decide what ought to be done.  But in at least a few situations, the human-assisted decisions haven't been all they should be. 

 

For example, an investigation by a safety agency found that a remotely-operated Waymo vehicle illegally passed a school bus where children were boarding.  Apparently, the Waymo remote assistance operators do not take direct control of the vehicle, but merely guide it in making its own decisions.

 

Tesla's process of remote assistance seems to be more assertive.  In responding to Sen. Markey's inquiry, Tesla revealed that in rare cases, their remote assistance operators can take direct control of the vehicle, but its speed is limited to either 2 or 10 MPH, depending on whether the internal automated system grants direct control to the human controller.

 

That makes it sound like the final decision is up to the machine, not the human.  Whatever the details, the fact that human remote assistants are sometimes involved in driverless car control adds another layer to the already complex mixture of responsibilities involved. 

 

Both Tesla and Waymo emphasized that the remote assistants do not sit there watching each car passively.  For one thing, there are too many cars to monitor.  Waymo revealed that by spreading the work among staff in the U. S. and the Philippines, they can maintain about 70 remote assistance operators on duty at any given time.  Tesla's operators are all in the U. S., and presumably work day and night shifts.  They all have valid U. S. driver licenses for at least three years prior to employment, for what that is worth. 

 

The information that all companies queried refused to reveal was how often it happens that remote-assistance operators intervene in a vehicle's operation.  Perhaps justifiably, they claim that is a trade secret that might compromise their competitive position.  My guess is that they have a pretty good idea of what the competition is doing, but they don't want the competition to know that.  Until it becomes an issue in a court case, then, we'll just have to guess at whether the driverless vehicle you see cruising down the street is truly on its own, or is being supervised from Palo Alto or Manila.

 

The mixing of autonomous and human control is going to become more urgent in the future as more and more systems become capable of running themselves.  Less than two weeks ago, on Apr. 14, Ukrainian President Volodymyr Zelenskyy announced that for the first time, his soldiers captured an enemy position using only remotely-controlled equipment, namely unmanned drones and "ground robotic systems," meaning armed robots. 

 

This historic achievement speaks volumes for the progress that Ukraine has made in developing agile, cost-effective military technology that augments their strength in human troop numbers.  After four years of war against a numerically superior foe, the Ukrainians have become world leaders in such technology, motivated by one of the strongest reasons around:  survival.  It is a truism in the history of technology that wars advance technical developments faster than peace does, and the Ukrainians have shown its validity once again.

 

For years, much ink has been spilled in engineering ethics journals about the dangers of autonomous warfare.  While this may not seem to have much to do with remote assistance for Waymo and Tesla, the two cases are extremes on a continuum.  In the case of peacetime vehicles, you want the machinery to do what it's supposed to do benignly without hurting anybody, but in cases where there is potential danger, you want the human backup person to intervene.  In the case of autonomous warfare, you want to damage and destroy the enemy and its resources, but you don't want the autonomous systems going haywire and attacking your own troops. 

 

Getting back to remotely-assisted civilian vehicles, it appears that the operating companies have found a comfortable medium between the extremes of completely autonomous operation—no humans involved anywhere along the line—and what would amount to a remote-control cabdriver, namely one remote assistant dedicated to each vehicle full-time.  Their comfort zone appears to be a mix of mostly automated control, but some manual control in emergencies. 

 

While it would be nice to know more details, in a free market it's up to the firms themselves to decide the optimum mix, and they seem to want to keep the details to themselves.  Until there's a major lawsuit in which the remote-assistant issue becomes a critical factor, we may not learn a lot more about it. 

 

The question of responsibility is fundamental in engineering ethics.  With the rise of all sorts of autonomous systems that are not quite autonomous, the question gets complicated.  The best we can do is to learn as much as we can about how things are actually done, so when problems arise the public and regulatory agencies are not kept in the dark about where responsibility may really lie. 

 

And whether it's a Waymo car or an armed robot capturing an enemy soldier, it's a little comforting to know that humans are still at the top of the command structure, however remotely.  Unless I'm the soldier—then it may not matter.

 

Sources:  Andrea Guzman's article "Self-driving cars still rely on humans—but how often remains unclear" appeared in the Austin American-Statesman on Apr. 8, 2026 at https://www.statesman.com/business/article/tesla-waymo-remote-operators-self-driving-cars-22192151.php.  I also referred to an article describing the Ukrainian remote-controlled capture of a Russian position at https://www.politico.eu/article/volodymyr-zelenskyy-robotic-systems-russia-army-positions-ukraine/. 

Chicago Objects to Driverless Cars

You know a technology's beyond the infancy stage when politicians start paying attention to it.  Driverless cars such as those being test-fielded by Waymo and other firms have upset a couple of Chicago aldermen to the extent that they have tried to enact a ban on them, saying they're too dangerous.  In an editorial as remarkable for its brevity as for its directness, the Chicago Sun-Times claims to take the side of the future in a piece titled "Driverless cars on the road to the future."

According to the Sun-Times, the aldermen—Ed Burke and Anthony Beale—are worried about jobs.  More perhaps than many other large U. S. cities, Chicago is a working-class town, and drivers of many kinds—cabbies, truck drivers, delivery people, airport personnel—make up a significant number of voters.  As such peoples' representatives, the aldermen elected by a particular district should take the concerns of their constituents seriously.  And if driverless cars threaten jobs, well, trying to do something about it is within the rights of a reasonable politician.  There's a whiff of hypocrisy in claiming safety concerns about a matter that's really more about jobs, but no more than usual in today's political environment.

The editorial writers take the side of Illinois's Governor Bruce Rauner, who may sign a bill that would prohibit Chicago and other cities from enacting a ban on driverless cars.  Conflicts between municipalities and state governments seem to be cropping up more frequently these days, typically with big cities taking more progressive positions and getting reined in by more conservative state legislatures and governors (Rauner is a Republican). 

If driverless cars become a significant percentage of cars on the road, that will mark one of the biggest technological changes in transportation since the introduction of the automobile.  We've been told that it will come with tremendous advantages compared to today's status quo:  lower accident rates, less traffic congestion, and the freedom to use your commute time for things other than steering and using the brake pedal a lot.  And probably the most visible downside is what it will do to the job market for paid drivers.  Every car on the road that used to need a driver but doesn't anymore represents a potential lower- to middle-class job that becomes history. 

The basis on which the editorial favors driverless cars is what historians call the Whig theory of history.  This is the idea that the farther back you go, the worse things were, and that human history is an unbroken series of triumphs over ignorance and primitive ways that will eventually issue in Paradise on earth.  After the horrors of the twentieth century (World War II, to name one), anyone who gives the matter a moment's thought will begin to see holes in the Whig theory.  But it's been around so long that it has become a kind of cliché idea that some writers spout automatically. 

The editorial writers claim that Chicago will become a "cow town" if it doesn't accept driverless cars, and they cite NASA's use of computers to get a man to the moon as an example of how computerized transportation is a good idea.  Well, these are not so much arguments as they are assertions and bad analogies.  The editorial winds up by saying no one banned Model Ts to protect jobs for blacksmiths, and "Old occupations may fade, but new ones come along."  The overall thrust of the article is basically to say, "Deal with it, and don't do something stupid that will make Chicago look like some kind of backward provincial hick town."

Beneath this rather trivial discussion are a number of serious questions.  What role should government play in the deployment of driverless cars?  Should their regulation be at a local, regional, state, or national level, or some combination of the preceding?  What, if anything, should be done to protect the jobs of people whose livelihood is threatened by the advent of driverless cars?  How can we get from the level of automotive safety we have now to a better one by implementing driverless cars, without running into some emergent large-number problem that could cause a significant increase in serious accidents, injuries, and deaths?  Almost none of these questions were addressed by the editorial writers, but if they were operating under a house rule that prohibits editorials of much more than 300 words, well, there's not a lot you can say in 300 words. 

By this point you may have the impression that I favor a ban on driverless cars.  I don't favor a ban, and I don't favor a law against a ban.  What I favor is a serious, in-depth discussion of the questions regarding driverless cars, and at least in the case of this editorial by one of the nation's major newspapers in a city of 2.7 million people, I don't see signs of that. 

There are many signs that the elusive thing called unity is on the decline in this country.  The degree to which citizens trust government to do the right thing has fallen precipitously compared to where it was several decades ago.  And people don't trust a system that they don't understand or feel that they cannot influence when it allows or encourages things that can cause them harm. 

Perhaps the two Chicago aldermen responded to their constituents in the wrong way, but at least they saw a genuine threat to jobs and went about trying to do something about it in response.  Back when the average person could read things that required more than one step of logic to understand, there were typically a few local newspapers to read in any major city, two or three TV and radio networks, and maybe the newsreel at the movie house.  That was it, as far as finding out what was going on in the world, and as a result, those who operated the media took care to use it with a reasonable degree of responsibility, and something like rational debate about great public matters of interest could be carried on.

But now, the media have fractured into a million tweets, Instagrams, and other detritus of electronic communications, most of which are too short to convey anything more than a burst of emotion.  The Chicago paper's editorial is only 300 words long because they probably know from experience that people won't read 1000-word editorials anymore, if they read anything at all.  And talking heads in video clips are pushing out text-based media of all kinds anyway.

I hope we as a nation, and the city of Chicago in particular, both reach a beneficial accommodation to the advent of driverless vehicles that will benefit most people while injuring as few as possible.  But to get there in a way that makes people feel included, the discussion will have to be at a higher level than the Sun-Times has set for us.

-->

Sources:  The editorial titled "Driverless cars on the road to the future" appeared on the Chicago Sun-Times website on Aug. 25, 2017 at http://chicago.suntimes.com/opinion/editorial-driverless-cars-on-the-road-to-the-future/.