Drunk driving might be eradicated with true autonomous cars. Meanwhile, Level 3 semi-autonomous cars might make it worse.
In the United States, approximately 11,000 people will get killed due to drunk driving this year. That’s 30 people, human beings, adults and children, male and female, killed each day of the year.
In terms of the number of people injured due to drunk driving, it’s hard to gauge, but some put the number around 750,000 harmed souls each year. Research suggests Americans are apt to drink and drive quite frequently, with an estimated 121 million such incidents annually. What can be done about this proneness to drunken-related driving and the resulting calamities?
An oft-stated justification for driverless autonomous cars is their potential to bring a precipitous drop in the rate of drunk driving-related deaths and injuries. Some pundits claim the number should become zero, namely zero injuries and deaths that are drunk driving-related, because truly autonomous cars won’t have human drivers, thus there’s zero chance of drunk driving.
This claim might have validity except for the fact that we aren’t going to suddenly have an overnight situation of all driverless autonomous cars and no human-driven cars on our roadways. The reality is that with 250 million conventional cars in the hands of Americans today, it will be many years before the emergence of driverless cars will overtake the stock of conventional cars, once autonomous cars start to become practical and available for use.
We need to grapple with the notion that the advent of autonomous cars will be gradual and slowly enter into the mainstream of car traffic.
No one even can say whether it will really be the case that ultimately we’d have only and exclusively driverless cars on our streets. Unless there is an outright ban on conventional cars, the odds are that some subset of the populace will cling to wanting to drive a car themselves and not have the AI be the only driver in-town. Thus, it keeps the door open to drunk driving.
That being said, if the amount of human driving gets smaller and smaller over time (being essentially stepwise replaced by autonomous driving), the volume of possible drunk drivers should be shrinking too.
And, the AI of driverless cars will presumably be shaped-up to deal with suspected drunk drivers that are encountered on the highways and byways. If an AI-run autonomous car spots driving behavior of other nearby traffic that resembles drunk driving, the autonomous car can seek to avoid getting entangled in the potential drunken driving act, along with warning other nearby driverless cars via V2V (vehicle-to-vehicle) electronic communications. This might reduce the odds of the drunk driver injuring or killing others.
In theory, an autonomous car that has detected a potential human-driven drunk-driving instance might alert the authorities, generating an urgent electronic communique to 911. Not everyone though is going to be okay with that approach. It would be like having a mobile system that is roaming around aka Big Brother and tattling on other cars. Yes, the AI might be right in terms of gauging and reporting on a drunk driver, but it could also be wrong and falsely accusing a human driver of something they are innocent of.
So far in this discussion, the focus has been on driverless autonomous cars.
For truly driverless cars, ones that are considered at Level 4 and Level 5 of automated driving, there isn’t a human driver and the AI is doing all of the driving. That’s a pretty clear-cut circumstance of ensuring that there is not a human driver at the wheel that might be DUI and veering towards disaster.
Meanwhile, on our roads today, the conventional cars of a Level 2 capability are starting to be outdated by the recent introduction of more automated ADAS (Advanced Driver Assistance Systems) in Level 3 cars.
Unfortunately, soon-to-emerge Level 3 semi-autonomous cars are not only likely to allow drunk driving to continue, but there’s also a possibility they could exacerbate drunk driving.
Let me explain why.
Seemingly A Paradox Of Level 3 And Drunk Driving
You might be puzzled that somehow drunk driving might actually worsen when there is increased automaton in our cars as Level 3 semi-autonomous vehicles gain popularity.
Shouldn’t better automation mean that we’ll have fewer injuries and deaths from this particular source? It stands to reason that theoretically it should. Here’s a rather poignant example of why it might worsen with Level 3 cars.
A reporter was recently interviewing General Motors CEO Mary Barra while the two of them were in one of GM’s semi-autonomous cars on a closed track. The reporter tried out some of the features that included letting the automation drive the car, even though the reporter was still considered the true driver and retained responsibility for the driving act per se.
At one point, the reporter turned to Mary and asked if the automation was essentially a designated driver that could take over the driving in case the human driver was drunk. No doubt many people harbor such a belief.
Indeed, overall, I’ve predicted that the Level 3 semi-autonomous cars are going to lull human drivers into falsely believing that the automation is capable in ways that it most certainly is not capable. Humans are humans, and if they start to rely upon the automation, you can be assured that they will fall further and further into the mental trap that the car can drive itself, even though at Level 3 it cannot do so.
Essentially, Level 3 has all of the trappings that seem like it can drive the car.
Some of you are saying that people will figure out the boundaries of what their Level 3 car can and cannot do. People are smart, they can get the drift of things. They will reason about their Level 3 cars and not allow complacency to overcome their judgment.
Besides my perhaps pessimistic view that I don’t have the same faith in humanity as the aforementioned notion, let’s also add something else into the equation. Let’s add being drunk.
Even granting that a completely sober person will be a good driver in a Level 3 car, and won’t let their attention drift from the driving task, which is a big if, suppose that person is intoxicated.
Do you really believe that an intoxicated person won’t assume that their Level 3 car can become their designated driver?
How This phenomenon Increases Drunk Driving
The likelihood that people will not only make the mental mistake of assuming their Level 3 cars can handle the driving for them in everyday situations when the human driver is sober, it will also and especially happen when those human drivers are drunk.
Someone is at their local bar and celebrating a promotion that they had gotten at work that day. Alcohol is flowing. After becoming tipsy, the celebrant wanders out to their Level 3 car in the parking lot.
With a conventional car, the person might have second thoughts about getting behind the wheel. In this case, given the staying-in-lane capabilities of the automation, the lane-changing features, and so on, the driver “decides” that whatever impairment they have will undoubtedly be handled by the Level 3 ADAS capabilities.
It would be an easy mental judgment to make, particularly when you’re sloshed.
Overall, Level 3 will create a false impression for those in a drunken state of mind that the automation will be their designated driver, making up for whatever the driver might have difficulties handling.
Some of you might suggest that in some respects this is only a partially valid notion. Indeed, there’s a chance the automation will warn the drunken driver that they’re weaving out of their lane. That’s good.
The hard thing to macroscopically calculate will be how much the Level 3 automation will be able to aid in keeping a drunk driver from getting into a disaster versus how much it will entice drunk drivers to try driving when they shouldn’t be on the road at all.
For every instance of the Level 3 serving as an on-board safety cushion to detect and warn when a driver is driving erratically, how many instances will we have of drunken drivers who purposely drove because they assumed the Level 3 could save them from their own drunkenness?
I’d tend to wager that the Level 3 won’t be sufficient to prevent drunk-driving injuries and deaths, in spite of its warning capabilities and in some cases its ability to undertake simplistic driving chores. I’d double-down on that wager and assert that Level 3 will be so alluring that it will encourage people who wouldn’t have considered driving drunk to attempt to do so.
In short, you’ve got people who would drink and drive no matter whether they have a conventional car or a Level 3, and they’ll still be getting sadly onto our roads. Level 3 won’t be sufficient to overcome their drunken driving antics and they’ll still on-the-whole commit drunk driving incidents.
Plus, you’ll now have an added contingent of the populace who otherwise would have been more circumspect about driving drunk who will mistakenly get behind the wheel under the mistaken and drunken mindset that their Level 3 car will save them.
As a result, incidents of drunk driving would rise–exacerbating injuries and deaths.
On the path toward true autonomous cars, we’re going to have semi-autonomous cars that are neither here nor there in terms of the muddying of what is automated versus what is not automated for the driving task. In the gray area of Level 3, people will assume it can be a co-sharing buddy that will keep the human driver from getting into trouble.
Drivers might be tempted to drive when they should not be driving. After all these years of trying to drum away at `don’t drink and drive,’ a segment of society will believe this mantra is no longer applicable because they have a Level 3 car.
Will we need to put breath analyzers on all Level 3 cars to try and ensure that people don’t let themselves become wayward and aim toward driving when they should not be driving?
If the use of breath analyzers seems over-the-top to you, perhaps Level 3 cars could try to use some other means to ferret out a drunk in the driver’s seat. With facial recognition and a camera pointed at the head of the driver, the automation might be able to detect facial expressions indicative of being intoxicated.
Furthermore, Level 3 could also potentially use voice systems such as the Alexa and Siri kinds of Natural Language Processing (NLP)systems that are becoming ubiquitous. Via the NLP, the Level 3 car could ask questions of the driver, and by analyzing their voice patterns and what they say, it is conceivable that the automation could ascertain the chances of the driver being drunk.
Will though people be accepting of having their own cars trying to challenge them as to whether they are sober enough to drive?
It seems hard to believe that people en masse will agree to such technology.
Maybe the tech would be offered on Level 3 cars and people could choose to switch it on or off.
Of course, you could argue that those that would be prone to drink and drive are probably the ones that would permanently switch off those features and thus it would only be those that wouldn’t drive drunk that would end-up having them switched on, pretty much defeating the purpose of those “early detection” systems for preventing drunk drivers from driving ahead.
Perhaps people will change their ways and with the advent of Level 3 we’ll see that drunk driving declines. I’m not holding my breath about this possibility and suggest that you’d need to be rather optimistic about human behavior, the very same human behavior that already kills 11,000 people per year in the United States and injures another several hundred thousand.