Nonprofit, nonpartisan journalism. Supported by readers.


Community Voices features opinion pieces from a wide variety of authors and perspectives. (Submission Guidelines)

Self-delusion and the self-driving car

Although self-driving car development is advancing rapidly, society has yet to seriously grapple with its consequences, including issues of consumer safety and personal liability.

The premise is simple: Enter the vehicle, select the destination, and sit back while the car does the driving for you. 360-degree cameras and sensors coupled with artificial intelligence work to effortlessly pilot you across town.

Photo by Adrian Danciu
Matt Ehling

The car’s appeal is that made time and again by the high-tech industry: Simplify life by letting complex technology do the heavy lifting. Enter a “frictionless” experience devoid of hassle. The tech press has been rife with stories touting the presumed benefits. “I never want to drive again,” wrote Yahoo’s Jillian D’Onfro after a recent Google press event that highlighted the company’s self-driving car program. “It just seems so dangerous. I’d rather put my trust in lasers and cameras.”

Google is currently testing self-driving (SD) cars on the streets of Mountain View, California. These retrofitted vehicles are accompanied by human drivers in case of malfunction or failure. However, SD vehicles are also being used on Google’s corporate campus that lack any human controls — even components as basic as a steering wheel. The company’s long-term vision sees the human driver as an engineering liability that needs to be excluded from its operational systems. “We’ve been working,” Google’s Chris Urmson has said, on “vehicles that can shoulder the entire burden of driving.”

While Google has garnered much of the publicity surrounding the development of self-driving cars, the presumed market potential is such that virtually every major auto manufacturer also has an SD car project under way. Daimler, for instance, is rushing to market with a semi-autonomous over-the-road truck — a precursor to the fully driverless vehicles of the future.

Although SD development is advancing rapidly, society has yet to seriously grapple with its consequences, including issues of consumer safety and personal liability. Beyond these policy-level considerations, SD technology poses more far-reaching and fundamental conundrums. The widespread belief that SD cars and other autonomous technologies are both inevitable and beneficial provides a window into the broader set of assumptions that dominate our present. It also illuminates how those same assumptions shield us from understanding what is truly happening to our world.

Losing our ability to make tech value judgments

The move to embrace SD cars (and other autonomous technologies) reveals a widely held premise that technology simply happens to us, and we must acquiesce to it. In this worldview, technology leads cultural development, and society trails behind, sorting out the particulars as it can. This has led, for instance, to a rush to mainstream drone aircraft, without any hard examination of the safety, privacy and liability trade-offs that the technology has wrought. The FAA has chronicled scores of near-misses between drones and commercial airliners, including one near-catastrophe over Tallahassee, Florida, in 2014. Despite this, the agency was tasked by Congress (after heavy industry lobbying) to implement rules for widespread commercial adoption as quickly as possible.

Like drones, SD cars may be thrust into mainstream usage along a similar path, without sufficient scrutiny of their many possible liabilities. These liabilities include the potential for hackers to remotely commandeer these Internet-connected vehicles, or to strip them of sensitive user data. Such threats are not theoretical. In a July article, Andy Greenberg of WIRED magazine demonstrated how a net-enabled Jeep Cherokee could be accessed from a laptop computer, to the point where the vehicle’s air conditioning, wipers and transmission were all compromised and controlled.

Similarly, the vulnerability of net-connected financial transaction systems has been well documented, and examples of security breaches continue to accrue in both the public and private sectors alike. Despite this, our society has not rejected such systems, and instead appears to be doubling-down on their use, with new electronic payment modalities coming online at a blistering pace. All of these trends illustrate a growing societal inability to make discerning judgments about the value and ramifications of the technological platforms that are being heaved at us. In this, SD cars are but the latest example of an accelerating trend.

Diminishing human skill and control

The going assumption that all technology is inevitable (and must be widely adopted) masks the trade-offs in individual autonomy and skill-development that are being made as increasingly automated technology is thrust upon us. We can look to Nicholas Carr’s 2014 book “The Glass Cage” for an in-depth examination of the impact of automation on human society, and the dangers of losing manual dexterity and cognitive development to automated processes. From spell-check degrading memorized language capabilities to (in one sobering incident) airline pilots who were unable to take manual control of an airliner after the auto-pilot disconnected, Carr chronicles the dangers of humanity sinking into an automated technological cocoon.

Of equal seriousness is the issue of reducing human agency, in addition to the loss of human skills. The conceit of automated technology is that it will free users to perform other tasks, and to better use their time. The reality is that those users will have to give up privacy and individual control in the exchange. For instance, SD cars would — by necessity — be linked to detailed databases through Internet connectivity, allowing their on-board computers to take advantage of repositories of driving scenarios, maps, road conditions and so forth. Vehicle-to-vehicle communication would place the cars in constant communication with other nearby “smart” vehicles.” Control, then, would flow not from the driver of the vehicle acting on his or her own volition, but from a corporate-owned network that would operate the vehicle, gather user data and archive the results for future “optimization.” For the benefit of being able to wrap up the day’s work while commuting, the resulting trade-off would be the ceding of anonymity and personal agency to a distant corporate entity. With the advent of consumer robotics and digital assistants, even greater trade-offs are in the offing.

There was a time when such trade-offs would have been un-recognizable to American culture — from the colonial settlers to the free-ranging beatniks of Kerouac’s “On the Road.” Since then, we have undergone a precipitous cultural shift, from an active society that strove to overcome significant obstacles in the real world, to an increasingly passive one that turns to technological intermediaries to conduct its affairs. The result has been a slow degradation of our vigor and skills.

The automation of daily life is an emerging phenomenon that is running parallel to the increasing automation of the workplace. As marketable skills are lost to smart machines (even skills as basic as driving a car) the economic consequences will be considerable. Today, millions of people depend on driving to make a living, including more than 3 million over-the-road truckers. Their disappearance from the salary roles will be not be insubstantial.

Individual-centric to network-centric world

SD cars have caused a stir in urban planning circles due to their anticipated impact on travel patterns and urban design. Many of the “efficiencies” of the SD concept are predicated on the idea that automotive companies will eventually stop selling cars to individuals, and will instead provide “transportation solutions” — networks of autonomous vehicles that circulate on an ongoing basis, waiting for app-based calls for service. Think of Uber, but without a human driver. (Indeed, this is exactly Uber’s long-term business strategy).

This “network-centric” shift represents a fundamental change to our nation’s economic model. The move away from the purchase of cars by individuals, to the “per-ride” use of SD vehicles owned by a centralized entity would be another step in the transformation of America from an ownership society to a rental society. This cultural change has been spurred by several macro-economic factors (including rising income disparity), and has been encouraged by the high-tech industry itself.

While the tech sector has been an active marketer of individual-sale items such as smartphones, the industry’s underlying strategy has been to leverage those devices as a way to encourage the rental of third-party services such as data-sharing plans and apps. In the business world, the tech industry has pushed cloud computer storage in lieu of the purchase of discreet hard drives for the maintenance of business data. Silicon Valley has also used the rental model to actively undercut markets for tangible goods (such as books) by making electronic versions those items available on a per-use, digital-only basis. Such a model ultimately places the terms of access squarely in the provider’s corner, allowing businesses to totally control the availability of goods and cultural artifacts, and to eliminate any after-market sales.

This approach is now bleeding over into established blue-chip industries like auto manufacturing. We can look, for instance, to General Motors, which recently adopted arguments that consumers do not actually own the cars they’ve purchased, but merely lease them, due to the presence of proprietary computer software in the cars themselves. SD-driven transportation networks would further expand upon this logic, entirely eliminating physical ownership at the individual level. Continuing to move in this direction will produce broadly felt economic consequences.

The production and sale of hard goods results in individual asset accumulation, as well as aftermarket sales and services that are widely distributed. In the automotive context, think of the broad-based economic impact of corner repair shops and family user car dealerships. Conversely, the rental model results in an accumulation of assets and a centralizing of capital by large-scale entities. For all its early talk of “democratizing” processes via the Internet, Silicon Valley has in many ways been a major driver of centralized control in our society. For instance, as Facebook has become a digital repository and delivery system for personal photographs, it has gutted the market for hard photographic goods, displacing thousands of small business photo lab jobs, while employing fewer than 10,000 people itself. It has also subjected all of its users to continuous, centralized surveillance. In this same way, SD car networks would create yet another consolidation of capital, another round of job destruction, and would impose network control in place of real individual autonomy.

Resource scarcity

Undergirding this entire discussion, of course, are questions of resources and their availability. As with the larger enterprise of industrial society, our assumptions about an ever-rising trajectory of technological progress rest upon an underlying premise that resources are limitless. 

SD technology is computer-intensive, with electronic components that would be forged from so-called “rare earth” elements (REE), including gallium, tantalum and terbium. The computer gadget-based economy that has mushroomed over the past two decades has increased the worldwide consumption of REEs in a similar, exponential pattern.

The “rarity” of rare earth elements stems from the fact that they generally occur in fairly diffuse concentrations, making commercial recovery difficult and costly. The burgeoning global demand for such materials, coupled with political factors, has already caused supply chain issues in the recent past. Since the late 1980s, China has been the dominant global producer of REEs, as it was able to bring elements to market largely without regard for environmental consequences. China’s pullback in rare earth exports in 2010 caused prices to spike worldwide, and put the question of REE scarcity squarely on the table. In his 2012 book “The Race For What’s Left,” author Michael Klare chronicled the difficulties of extracting the world’s known REEs, noting that in addition to being naturally diffuse, the elements are frequently bundled up with radioactive isotopes such as uranium, thus requiring costly mitigation and disposal as part of the mining process.

Replacing much of the first world’s vehicle fleet with SD cars, on top of the burgeoning use of computing power in the “Internet of Things” and omnipresent portable electronics may be economically impossible at some point, given the finite amounts of commercially extractable REEs. Ultimately, society will be forced to make resource-based choices about computerizing and automating aspects of our daily life — particularly when the same pool of REE resources are required for other, competing industrial uses that are also on an upward trajectory. These include the use of REEs in energy efficient and “green” technologies, such as wind turbines, LED lighting, and hybrid cars.

While the transition to autonomous technology will have significant cultural and economic impacts, the real-world resource constraints that accompany our society’s push into the “frictionless” future of smart machines may be the factor that matters most in the end.

The future depends on ourselves, not on our machines

All told, human beings have driving motorized vehicles for over 100 years, and have piloted horse-drawn equipment for centuries more. Now, however, Silicon Valley executives tell us that such activities are far too dangerous for people, and should be ceded to smart machines for our own good. In the view of Tesla CEO Elon Musk, cars with human drivers may eventually be made illegal because of human fallibility.

The tech industry’s push for an automatic future for cars is symptomatic of its broader attempt to create a world where humans are increasingly “designed out” of the processes that run our society — everything from stocking shelves to financial trading. Technical refinement and complexity, we are told, can rise above imperfect human activity, and can improve our ailing world by removing us from its management.

Such an overly complex future is not a solution, however, but a trap — culturally damaging in the short term, and materially untenable in the long term. It is a trap that can only be avoided by awakening from our machine-induced slumber and viewing ourselves — and our world — anew.

Matt Ehling is a St. Paul-based writer and media producer who is active in government transparency and accountability efforts.

Want to add your voice?

If you’re interested in joining the discussion, add your voice to the Comment section below — or consider writing a letter or a longer-form Community Voices commentary. (For more information about Community Voices, email Susan Albright at

You can also learn about all our free newsletter options.

Comments (9)

  1. Submitted by Adam Miller on 10/23/2015 - 10:09 am.

    What is truly happening to our world

    Is that humans are dying on our roads at appalling rates, in the U.S. much higher than the rest of the developed world. Humans are terrible and unsafe at operating motor vehicles. The possibility of handing that particular task to a machine that can do it much more safely is a really poor jumping off point for a lament about lost autonomy. Here, autonomy kills.

  2. Submitted by William Lindeke on 10/23/2015 - 11:27 am.

    great piece

    Matt: I really appreciate your article here. I share much of your skepticism about liability and glitches. See the recent Tesla autopilot shenanigans that are already happening on our US roads ( What happens when something breaks, goes wrong, etc.?

    The reason that urban planners are so intrigued by the concept, though, is because if might offer a potential transformation of the problematic and extremely persistent system of automobility that dominates our cities and lifestyles. The potential to delink the personal automobile from personal transportation could be revolutionary for many key facets of our lives, from housing to health to local government finance.

    Meanwhile, however, see the other article on car technology in today’s Minnpost: Car manufacturers and phone companies continue to make failed promises about wedding technology to driving. These kinds of things (checking Facebook via voice command) are exactly what I think you’re talking about when you say we’re losing the ability to discern what technology is important. You need only read the last texts of drivers who have died while texting to see how meaningless so much of our technological distraction can be

    The simple fact: safety in cities depends on drivers slowing down and focusing their attention. I’m really worried we’re heading in the other direction.

  3. Submitted by Frank Phelan on 10/23/2015 - 12:54 pm.

    Franken Foods

    I’d like to know what the writer thinks about GMO foods, since many like to poo-poo anyone who questions the benefits of them. For years infant formula manufacturers have sold parents adulterated cows milk, promoting it as “just as good as” mother’s milk, the only food designed by mature expressly for human consumption. (Cow’s milk is designed for… baby cows.) Time and again they have found they have fallen short and come up with”new and improved” varieties.

    Some distrust all science, some worship at it’s altar.

  4. Submitted by Wayne Coppock on 10/23/2015 - 01:26 pm.

    Economic shifts happen

    The author complains about the potential to upend a huge portion of the economy dependent on driving, but is this really anything new? Does anyone cry for the stable boys of the past that tended to the huge numbers of urban horses that disappeared? Does anyone lament the end of railroads as the dominant method of moving goods? Do we long for the days of steam ships and canals?

    This kind of hemming and hawing is nothing new in the lens of history and, as with the same sets of concerns and complaints in the past, it will be a footnote at best. The huge economic efficiencies to be gained by a shift to self-driving vehicles is enough for many, but the *safety* improvements really put it over the top. Human beings are terrible drivers. Technological advancement has already done plenty to reduce their skill and attention to driving and because of environment we built everyone is expected to drive, including those who are really not fit to do so. But because we’ve created this world where not being allowed to drive is tantamount to excluding someone from society, we force nearly everyone to participate in this system regardless of whether they want to or should.

    I’d personally much prefer a huge investment in public transit over self-driving cars, but my wishes don’t drive market forces and self driving vehicles seem to be the way most investment is going. It probably has something to do with cars being in the domain of private investment and public transit being in the dysfunctional public funding domain (never mind that the roads required for car use are also funded publicly, they are treated differently by governments in a clear double-standard). But if self-driving cars is going to be the transportation solution we’re going to move towards as a society, we need to make sure the solution actually works for everyone instead of just crying about how we don’t like it or it’s different and we’re afraid of change. I don’t see any suggestions for ways to direct the development and legislation around these in the article, just complaining and hand waiving about how the author doesn’t like the way things are headed. But he and his ilk won’t stop or even slow down the coming changes, and this kind of neo-luddite attitude has no place in the modern world.

  5. Submitted by Rick Prescott on 10/23/2015 - 03:46 pm.

    Fear, and More Fear

    The author of this post betrays what’s really behind his views when he uses the absurd phrase, “The computer gadget-based economy…”

    Self-driving cars will add to personal freedoms far beyond just eliminating (yes, eliminating) the possibility of dying in a car crash. Freedom from vehicle ownership and maintenance is already sought by many people. Managing vehicle emissions as part of a fleet has the potential to drastically change the carbon footprint of transportation. Public transportation will be cast in a completely new — and potentially much more cost effective light, while also adding a level of flexibility for the poor that is unheard of in a bus/train world.

    Genuine advances in technology have always had to face irrational resistance. And despite a fine clarity in his writing, this author stands on the same ground as the horse-and-buggy people who resisted automobiles.

    Of course, there will be problems, just as there are in any major cultural/technological shift, and privacy and security are very real concerns which need to be addressed. Vigilance is a necessity, as it always is in moments of sweeping change. But the potential advantages of self-driving cars far outweigh the potential downsides (the same being true for automated flight, incidentally, with pilot error being by far the single most common cause of aviation accidents). Imagine living a world in which there is simply NO SUCH THING as DWI. Only in an absurd world would such genuine advancements in safety and convenience be rejected because it might displace jobs, or comes with other minor caveats.

    One suspects that this author may have resisted having a computer on his desk at one time. But that technology, initially terrifying, intimidating, and disrupting to so many people, now makes this very exchange of ideas and opinions possible. As such, any arguments against technological advances which are based in a fear of the unknown must be discarded. They simply cannot and should not prevail.

  6. Submitted by Dean Knudson on 10/26/2015 - 07:29 am.

    Data are the data

    Not to toss a wet blanket on the argument…but…let’s use science.

    The argument that self driving cars are safer than human driven cars is testable. Implement self driving cars in a limited fashion, and compare them statistically against a scientifically selected sample of human driven cars, in a fashion that can be replicated and published in appropriate peer reviewed journals. After a number of successfully replicated studies, you would have an answer with a confidence interval of better than 95%.

    One might find that implementation of self driving cars would save a substantial number of lives. Right now about 30,000 people die in car accidents yearly, and a majority of those deaths involve human error, apparently. Not surprising, we did not evolve with cars, and are not always well suited to operate them with fail-safe precision and constant and invariable attention. We are not wired to do that.

    One might find that self driving cars would save 15,000 lives a year. At that point, with a confidence interval of 95% certainty, one could safely say that NOT implementing a universal system of self driving cars would condemn 15,000 people to death each year.

    The law of large numbers still applies, however. Self driving cars WILL kill people. Has to happen. We then would be in a politically difficult position….we’d have to ignore anecdotes and stories, and adhere to science, to save those 15,000 people yearly, and that would be tough to do, as the small number of deaths related to self driving cars would cascade out of cable TV daily.

  7. Submitted by Brad Koons on 10/27/2015 - 09:33 pm.

    Flawed Argument

    I have many objections to the premise of this article, but I’ll make two points around the following statement:

    “All told, human beings have [been] driving motorized vehicles for over 100 years, and have piloted horse-drawn equipment for centuries more. Now, however, Silicon Valley executives tell us that such activities are far too dangerous for people, and should be ceded to smart machines for our own good.”

    1. Humans are awful at driving cars. The worldwide annual death toll from human error and incapacitation (drugs, medical events) is known and real and overcomes an abstract argument about loss of privacy and hand-eye coordination.

    2. Horse-drawn equipment had animal intelligence built in. A drunk could get home safely without injuring himself or others because the horse would take over. Horses are fallible, but I’d take a car with the sensory system and intelligence of a horse over current cars.

    The only reason that motor vehicle deaths have declined over time is due to technology – seat belts, air bags, engineered crumple zones, rear backup cameras, etc. – versus continued improvement of human drivers.

    I welcome a future with automated cars. Furthermore, from a public policy standpoint I’m disappointed that the federal goverment hasn’t applied more focus on advancing this technology. The cost savings from a significant reduction in deaths and serious injuries (medical costs, productivity losses) and the more efficient use of existing road infrastructure are a compelling argument for public augmentation of market-driven R&D investment.

  8. Submitted by Per Inge Oestmoen on 02/11/2016 - 08:31 pm.

    We should reject self-driving cars without fail

    Do we really want self-driving cars?

    Will self-driving cars mean the end of highway fatalities?

    In the same way, one might ask:

    “Will the complete prohibition and confiscation of privately owned firearms mean the end of homicides by shooting?”

    And further:

    “Will the act of mandatory surveillance of every minute in our life mean the end of domestic violence, rape, public violence, homicides and theft?”

    If technology so permits, it is always possible to mandate ever more prohibitions, restrictions and limitations on practically everything that might conceivably pose a risk to humans – and then the question of whether the act will mean the end of phenomena which we fear or dislike will increasingly be a moot one.

    The point is that if the aim is “zero tolerance” towards traffic accidents and casualties, homicides or fatalities from children falling down from trees in the kindergarten, people will have to realize that those who give up more and more of their choices and possibilities in order to gain (an illusory) safety will lose both their freedoms and the safety. If self-driving cars are to function optimally the whole traffic must be structured around a complete prevention of people driving their own vehicles – including four-wheel cars and motorcycles.

    Thus, the discussion around self-driving cars is not about ending highway fatalities. It is about society’s capability of tolerating the presence of risks in everyday life. It is sometimes stated that self-driving cars are coming slowly but surely. That is wrong. The driverless cars, and the accompanying inevitable universal prohibition of manually driven cars in such a scenario, do not “come.” Rather, they are brought in by humans who are willing to sacrifize their abilities to drive a vehicle, their freedom to use their driver-controlled vehicle anytime, anywhere and with the use of their own senses and motor abilities.

    Yes, by all means manual driving with our own acquired abilities entails risks.

    To let children climb in trees entails risks.

    To permit people to own firearms entails risks.

    As a matter of fact, to permit people to procreate sexually also entails a lot of risks that might conceivably be ended if non-sexual reproduction became mandated by a government which had zero tolerance towards all the fatalities that result from the processes of conception, pregnancy and childbirth. Think about it; vast amounts of money would easily be saved if biological procreation and all the enormous hassle around it could be supplanted by hi-tech in
    laboratories. Yes, the conflicts between sexes could be brought to an end if the sexes themselves were eliminated. Orgasm can easily be achieved by artificially stimulating the pertinent parts of our nervous system, there could be an “orgasmatron” for that.

    Is this line of reasoning far-fetched? Hardly, it is no more far-fetched than it would be if someone argued for (or against) self-driving cars back in 1970 when 560 people lost their life in the road traffic in Norway. In 2012 with vastly better roads, but with 2.8 million cars and 5 000 000 – 5 million – inhabitants, a mere 154 people died (2012). That number is far lower than the annual number of people who fall prey to influenza in Norway. To put that in proper perspective, it is estimated that during the first decade of the 2000’s between 7000 and 8000 humans expired from their use of tobacco. Each year, there are approximately 42 000 deaths in Norway among 5 000 000 people. Of these, less than 200 die from traffic accidents. In 2015, only 125 were killed in a population of more than 5 000 000. That is hardly an enormous number. We need to put this in proper perspective.

    The force behind some people’s fascination with self-driving cars is precisely the same tendency as we see when children are denied their right to climb in trees because they might fall down and become injured, and when some people would like universal electronic surveillance by the state in order to put an end to crime.

    Every time, it is possible to defend a new restriction with the treacherous allure of putting an end to an unpleasant phenomenon or a danger – or some real or imaginary expense for that matter.

    Alas, maximum safety means minimum choice, minimum freedom. In other words the freedom, the choices and ultimately many of the things that enrich life and give us capabilities in our everyday life will be removed – but the complete safety will forever elude us in a living world. That is the way it is; like it or not.

    To conclude, the question is not whether one can end fatalities on the highway or for that matter anywhere else. The really important question is how far we are willing to go, and how many things we are wiling to deprive ourselves of in order to try to eliminate what we fear or dislike – or what is politically correct to remove at any given time.

    – Perhaps, just perhaps, it would be wiser to accept life’s risks.

Leave a Reply