By Dr. Lance B. Eliot, the AI Insider for AI Trends and a regular contributor
Do you care about self-driving cars? If you are an AI specialist, the odds are that you care to some degree, since the hottest AI-action is right now taking place in the autonomous vehicles arena. Salaries and bonuses are humongous for those AI developers that are aiming at the self-driving vehicles market. Researchers that once were hidden in the basements of university labs and toiled away on driverless modes of transportation are now sought to create what some believe will be the most disruptive technological breakthrough in modern times. Auto makers are scrambling and rushing to get into this gold rush. My prior columns have well documented the anticipated impacts on business, society, politics, and our way of living.
Does the general public care about self-driving cars? If so, what do they know about self-driving cars, or, rather what do they think that they know? Much of the public perception about self-driving cars is based on a combination of real news and fake news (see my column covering the fake news and its impacts). The public gets glimpses of what is happening with self-driving cars as fed to them by occasional coverage in the major media outlets. Remember the pretend game you did as a child involving closing your eyes and then randomly opening to see what is going on, and then trying to piece together the reality by those short snippets? That’s what the public is doing today.
Recent polls are trying figure out the nature of public opinion regarding self-driving cars. Seems like we see a new poll every week. In some cases, the poll sways the voting by portraying self-driving cars as either savior of goodness or as evil doomsayers. You can get the public to pretty much go in a particular direction by how you phrase the questions. It’s similar to polls done for political races. If I ask you whether you support James Smith for president, I’ve worded the question in a relatively neutral fashion. If I instead ask you whether you support a lying, cheating, horse stealing James Smith for president, I have loaded the question in a direction that is likely to get you to say no. I might instead ask whether you want the gallant, honorable, hero James Smith to be president, and you’ll shift toward saying yes. Of course, some people will already have a predetermined opinion regardless of how the question is worded, but those that are on-the-fence or not in the know can be maneuvered by how the question is phrased.
I tell you this aspect because I’d ask that you gauge the results of public surveys about self-driving cars with a grain of salt. A headline that announces the public is eager to have self-driving cars can be inherently biased by perhaps asking whether someone wants a self-driving car that will save lives and reduce pollutants. Who can disagree with that kind of question? Not many. On the other hand, if the survey asked whether you want self-driving cars that will endanger lives, drive recklessly, and start us toward a future of AI taking over the world, I am betting that most would insist they don’t want self-driving cars. Trickery by the designer of the survey is just as much a part of the survey as anything else that the survey purports to do or say.
Not only is the survey design crucial, it is also important to consider to whom the survey is administered. If I give a survey about self-driving cars to researchers and AI specialists, I’ll likely get a result that would differ from giving the same survey to teenagers in high school. There’s a famous case about survey selection problems that took place during the presidential election in 1936. A popular magazine called The Literary Digest sampled 2.4 million Americans to then predict which candidate would win the presidential vote. Turns out The Literary Digest prediction was wrong, and led to a shocking surprise since it was one of the largest sized samples ever done for a national poll. How did they get it wrong? They selected names by looking at their subscribers and telephone directories, primarily of which were higher income and white. This is known as selection bias. They also originally put together a list of 10 million names, but only those that actually responded to the survey were the 2.4 million people. This is a nonresponse bias that meant only those that did respond cast an opinion, while the millions of others that did not respond we have no idea what their opinion might be.
Be careful therefore in believing any self-driving car poll that you see. Some of the polls are done by professional survey experts that know how to properly design and conduct a survey, and for those polls I would gauge that we can generally accept their results. Other polls that are done ad hoc by someone that knows nothing about properly doing polls should be considered highly suspect. There are also some polls that want to get a particular answer, and so those pollsters will then purposely shape the questions and the selection of respondents to get the outcome they want. If you are an organization that wants to bash self-driving cars, you can easily create a poll that gets that kind of public perception results. If you are an organization that wants the public perception to seemingly be that self-driving cars are essential, you can get that response by how you develop the survey.
Let’s take a look at a specific poll that was recently undertaken and gained national and international attention and was undertaken by the American Automobile Association (AAA). The American Automobile Association is a well-known entity that provides various auto related services to consumers. The Triple-A, as it is informally called, consists of a federation of automobile clubs throughout North America and has a membership of around 50-60 million consumers, which is a hefty and very impressive number (the AAA has a lot of clout due to this count). Besides providing roadside assistance, the AAA also provides help in all forms of travel, car insurance, tourism info for places you might go visit, and the like. As you might guess, they have a big stake too in the advent of self-driving cars. Right now, most of their membership consists of drivers. This is a significant in that their bread and butter is due to drivers — keep in mind that if there are less people driving due to the arrival of self-driving cars, it could dramatically impact the AAA as an entity and its survival. Some believe that the AAA will have to radically transform itself to adjust to a predominantly passengers-only world of self-driving cars.
The AAA did a survey in January 2017 about self-driving cars and released Phase 2 of the results in March 2017. I am hoping you already are wondering how the questions were worded, and also wondering whom they opted to survey (if you’ve been paying attention to my raving and ranting in this piece). As forewarned, knowing those aspects is crucial to being able to interpret the results of a survey and also what kind of biases might have been purposely or accidentally embedded in the polling. In this case, they randomly selected landline phone numbers and also mobile cell phone numbers of 1,012 adults (ages 18 and over) that were living in the continental United States at the time of the polling. I won’t get into further details about their methodology here and urge you to take a look at the AAA published poll to see the other various assumptions and limitations.
The result that got the most headlines by the AAA consisted of the “finding” that 75% or three-quarters of United States drivers said they were afraid to ride in a self-driving car. We can readily quibble with this finding due to how they asked the question and also the sampling choices used. I am not going to fight that battle here, and instead pretend that the result is bona fide. As such, what does it mean?
Suppose you believe that self-driving cars are safe. Your interpretation of the aspect that 75% of drivers are fearful shows that the public is quite confused and they plainly don’t know what they are talking about. Perhaps they are naïve. Perhaps they think of self-driving cars like Frankenstein and so have a mistaken view of what self-driving cars are. Scientists often find that the public lags awareness of scientific breakthroughs and can be hesitant to embrace new innovations.
Suppose you believe that self-driving cars are not safe. Your interpretation of the 75% result is that thankfully American drivers are astute enough to know that they should be wary of self-driving cars. Whether these people knew what they were saying, or were just basing their opinion on a hunch, they got it right. This can be used to prod the self-driving car makers into being more focused on the safety aspects of self-driving cars. It also enables politicians to put in place regulations about self-driving cars, which they can say is being done due to overwhelming public opinion that self-driving cars are perceived not to be safe.
There are some additional results that you might find of interest, and that might also make your blood boil, depending upon your stance about self-driving cars. The poll indicated that women tended to be more afraid than men about riding in self-driving cars (85% of women versus 69% of men). Some might say that this is because women are likely to be more fearful of mechanical things (there’s a misogynist view for you!), while others might say that women are more enlightened and less willing to blindly believe that these self-driving cars are safe (men let their bravado do their thinking for them).
The survey also reported that if you divide the respondents into Millennials (ages 18-36), Generation X (ages 37-52), and Baby Boomers (ages 53-71), the group that was more fearful was Baby Boomers (85%), then Generation X (75%), and then Millennials (73%). We might use the politically Impolite aspects of ageism and declare that the oldies “don’t get” new technology, while the youngsters do. On the other hand, some might say that the elders have wisdom that they’ve seen many a technology that was over-hyped and turned out to be worse than we were being told.
If you are a self-driving car maker, and if you believe this poll, the point for you is that you need to do something to get the public to believe in self-driving cars. I suppose you could jack-up the belief in self-driving cars by getting some hot celebrity to go around in a self-driving car. You could do a marketing campaign about how unsafe human drivers are, and thus get the public to de facto assume that self-driving cars must be safer. And so on.
Another question asked by the poll involved whether human drivers feel safe around self-driving cars, when both are on the road together. Now, you would almost assume that if 75% don’t feel safe being in a self-driving car, they certainly would also not feel safe being near one on the road. In fact, you would probably assume that an even higher percentage would say that they don’t trust the self-driving cars to be mixed with the human drivers. Well, you’d be wrong. Turns out that only about half or 54% said they would feel less safe sharing the road with self-driving cars. I guess the logic must be that as a human driver in your own car that you can avoid the unsafe nature of the self-driving car nearby, but if you are actually in a self-driving car then you are stuck at the whim of the unsafe self-driving car. Or something like that.
The last item that I’ll cover about this poll consists of a question asked about how the public perceives the nature of the autonomous systems that would be on these self-driving cars. According to the poll, approximately 8 of 10 said that they want the autonomous systems to work in the same way (81% said so), irrespective of which car maker provides the features. This is not exclusively about self-driving cars per se, since the question defined the autonomous systems to include lane keeping, adaptive cruise control, self-parking, automatic emergency braking, and so on. It certainly makes sense that people would want these features to be consistent across car makers and models.
I’ve already written about the aspect that at first we’re going to see the car makers using their self-driving car capabilities as a means to differentiate themselves in the marketplace. No auto maker that has spent billions of dollars is going to like the idea that all self-driving cars do the same thing. Instead, they each want to have their own competitive advantage. For consumers, it will be the wild west of trying to figure out what one auto maker has versus another. Is the self-driving capabilities of a Ford any better or worse than those of a Toyota? And, how can the consumer know for sure, other than the massive advertising that will be done by the auto makers. Entities like the AAA will be looked upon by consumers to help them figure out what is the truth about the auto makers claims.
Overall, these polls are handy to try and ascertain what the public thinks about self-driving cars. Please don’t believe a poll simply because it seems to agree with your own opinion. Look under the hood, so to speak, and see what the survey design consisted of and how they respondents were selected and contacted. If the survey results disagree with your opinion, don’t immediately reject the survey, and instead take a closer look at how it was undertaken. You might know that there’s a famous line about statistics, which was uttered by British Prime Minister Benjamin Disraeli: “There are three kinds of lies: lies, damned lies, and statistics.” For those that believe in the future of self-driving cars, if we see polls that seem overly distorted and misrepresent self-driving cars, we’ll need to speak out. At the same time, we cannot discount public perceptions, even if it is poorly informed, as we presumably should help shoulder the responsibility to properly inform the public about the present and future of AI and self-driving cars. One thing for sure, we’re likely to see public perception rise and fall, one moment loving AI self-driving cars and the next trouncing them. It will be a roller coaster ride of public perception. Keep your eyes wide open, and make sure to buckle up!
This content is original to AI Trends.