Automakers, legislators, and others are excited about putting autonomous vehicles (AV’s) on the road as quickly as possible, and they cite many benefits. They’re pushing hard and investing a lot of money, but there are simple reasons why self-driving cars will not rule the road any time soon, no matter how beneficial they might be. For starters, some people will prefer to drive themselves or ride with a human driver. Inevitably some people will feel insecure riding in an AV and will refuse to ride in one, let alone buy one. While the numbers of those rejecting AV’s for such reasons may be small, they will sustain a demand for self-controlled vehicles to remain on the roads longer, presenting serious problems for AV operators. But that’s just the beginning – there are other major problems with AV’s that are not often mentioned in the media.
Hacking of AV’s is a serious risk. People knowledgeable in data and network security may be leery of AV’s because of the potential for hacking. What if hackers cracked the vehicle network and then sent people to different places based on their sex, race, or some other parameter, possibly categorizing people using their image as taken by in-car cameras (which will be attractive to fleet operators such as Uber)? What if hackers took over specific vehicles and used them to take unsuspecting people to out-of-the-way locations where they could be robbed? What if hackers caused traffic jams in precise locations and at chosen times in order to prevent police from stopping a bank robbery? Perhaps hackers will figure out how to get AV’s to pull aside and let them through in traffic jams, for example.
Autonomous vehicle computers will be susceptible to computer viruses. There is no computer that can’t be compromised given sufficient access, knowledge, and tools, and there are plenty of people who, for personal gain, prestige, or other motivations, will take great pleasure in showing off their programming skills by attacking AV computers and networks. Given the security-challenged nature of the modern internet, it is highly probable that computer viruses will soon appear to specifically target AV’s. Like all computer viruses these will mostly arrive via network connections (and interconnecting AV’s is a major goal and selling point for manufacturers), and virus impacts will range from minor nuisances to data theft, cyber vandalism, and network disruptions. Are you enjoying trying to protect your computers from viruses and malware? It will only get worse when you have a vehicle to worry about as well, and the future owners of fleets of AV’s, such as Uber and Lyft, may be vastly underestimating the future costs of vehicle and network security.
Members of society’s fringe groups may not embrace AV’s. One challenge will be winning over conspiracy theorists, some of whom will fear the government might use AV’s to control their movements and may fear that a government could take control of all the AV’s and then use them to (for example) selectively take political rivals and enemies to concentration camps. That is far-fetched, of course, but I know people (just a few) who would point out that jailing dissidents for speaking out takes place in a number of other countries, and while there is no precedent for this sort of thing in the U.S., people in other countries may have more to fear, and such events are not impossible (and may occur without people’s full understanding) in the U.S..
AV’s will attract attention, and some drivers will want to look at or “test” them. When I mentioned to friends that I was writing this article, one of them said he finds AV’s fascinating, and when he sees a car with rotating cameras on the top he purposefully moves his car close to it to see what it will do. I jokingly call that “playing chicken” with AV’s, and during the implementation phase for the technology there may be more of this kind of activity by human drivers than one might expect. I’m also waiting to hear of the first game of chicken between cars with adaptive cruise control/automatic braking.
Legal issues will need to be sorted out. As a recent USA Today article points out, in the U.S. an AV is regulated by the Department of Transportation, but as soon as a human touches the controls it is governed by the state in which it is being operated. This could turn into a legal nightmare until it is sorted out in the courts and legislatures. Such legal tangles may exist in other countries as well. If an AV is found at fault in an accident, the question remains as to who is liable, and none of the companies involved wants to be saddled with blame. Many new legal cases and legislative actions can be expected before the dust settles.
A major problem is the requirement for the “driver” to take control of their AV only rarely. In regular vehicles the driver must be vigilant, attentive to their driving, 100% of the time to preserve safety on the roads. Manufacturers have suggested AV’s will require a human to intervene only in certain driving circumstances, however, and this brings up a number of serious problems: how does the human attendant (driver?) know when to take control? How quickly will they need to take control, and can they do this quickly enough when their attention has been directed elsewhere for a period of time? What if the human has dozed off during hours of automated travel? Having people intervene under only certain circumstances will be a LOT more problematic than making them responsible for full-time monitoring, and many may be entirely unable to take control when it suddenly becomes necessary, resulting in many accidents and injuries. I predict a failure of the “driver attention on demand” concept.
Another worry – will autonomous vehicle programming bring on “the Terminator” scenario? Artificial intelligence has long been followed by concerns that computer systems would become sophisticated enough to become self-aware, and then could develop a concept of self-preservation and become defensive towards humans. This could possibly culminate in a full-blown scenario from the Terminator movies in which the machines are intent on wiping out all humans. Some limit this scenario to Ray Kurzweil’s “technological singularity” concept, but many others see no reason why thinking machines would have any allegiance to humans, and extend this into a variety of dystopian visions of the future, some specifically citing a possible “revolt of the machines”.
The human is still the most unpredictable variable. The mixing of AV’s and human-piloted vehicles on the roads seems like a significant risk and one wonders if the auto makers are thinking realistically about it. Humans are both creative and wildly complicated, and have an inherent day-to-day error rate of around one percent. That variability and unpredictability add up to a major problem for AV programmers. Past experience suggests that, no matter how many situations programmers anticipate and prepare for, humans will inevitably invent new scenarios that haven’t been considered and will result in crashes and other problems. Human variability cannot be overestimated, in my experience, and the mixing of AV’s with human-piloted vehicles presents problems that are far from being surmounted (and some may never be).
Autonomous vehicles have a longer way to go than the manufactures and media would like us to think. All of the above issues will at least slow the adoption of autonomous vehicles, and some may not be overcome for quite some time, if ever. Today vehicle manufacturers have sold the AV concept to legislators, who have frequently failed to understand similar complicated technical areas in the past, and the industry and its supporters are off and running, promoting the great value of changing to autonomous vehicles and planning for widespread adoption in just a few years. Unfortunately, the realities may be quite different. It is likely that a lot of wishful thinking is being promoted by manufacturers eager to sell more vehicles, as well as new-styled cab companies like Uber (eager to cut costs by getting rid of their communities of drivers). They are feverishly stoking the media buzz but, as so often occurs, achieving the visions they are promoting may present unanticipated problems and take a lot longer than they propose. My simple suggestion is, if you’re waiting for autonomous vehicles to relieve you of needing to drive, don’t hold your breath.
I have been watching trends and future technologies for years, but this is the first time I’ve assembled my concerns regarding autonomous vehicles in one place (and it feels good). Thanks for reading – Tim
Related reading: the Auto Extremist “Shiny Happy Riders in Zombie Cars”