The Sobering Aspects of Self-Driving Cars Contending With Drunk Drivers

Dr. Lance Eliot, AI Insider

Image for post
Image for post

(Ed. Note: For reader’s interested in Dr. Eliot’s ongoing business analyses about the advent of self-driving cars, see his online Forbes column:

A sobering statistic: Drunk drivers account for nearly one-third of all traffic-related deaths in the United States (per stats by the National Highway Traffic Safety Administration).

The rule-of-thumb is that there’s an alcohol driving related death every hour, based on averaging the number of such deaths over the course of a year.

You might not realize that annually there are more than 1 million drivers arrested for driving under the influence (DUI), which is an equally scary statistic (one can only wonder how many of those might have gotten into a deathly incident were they not arrested!).

Of course, the one million drivers only represent those that were actually arrested and so presumably there would be many more that didn’t get caught.

Some surveys indicate that there are perhaps more than 111 million instances of DUI “episodes” per year by U.S. adult drivers (this is based on self-reported indications by drivers).

Though those episodes might not include actual driving related deaths, they likely include a significant number of driving fender benders, car or pedestrian sideswiping, frightening near-misses, and other dangerous mishaps.

Overall, you would be wise to be on the watch for drunk drivers.

Driving Actions Of Drunk Drivers

Let’s consider the types of driving actions that could be an indicator of a drunken driver:

  • Driving too slowly for the roadway situation
  • Driving too fast for the roadway situation
  • Nearly hitting another car
  • Cutting off another car
  • Swerving across lanes needlessly
  • Straddling a lane without apparent cause
  • Taking wide turns rather than proper tight turns
  • Driving onto the wrong side of the road
  • Driving onto the shoulder of the road
  • Driving in an emergency lane
  • Nearly hitting a pedestrian, bicyclist, or motorcyclist
  • Being too close to the car ahead of it
  • Stopping when it seems unnecessary
  • Rolling past stop signs
  • Running a red light
  • Other

I realize that those driving actions could be accounted for by some other reason beyond just drunk driving.

As such, I am not saying that it is automatically an indicator of drunk driving if you happen to see a car do any of those specific actions.

You need to look at the context of the driving situation.

Overall, those types of driving acts are telltale clues that can be used to try and guess whether there might be a drunk driver involved. It is useful to try and assess whether a driver is a drunk driver, since it tends to suggest that they will be dangerous in their driving efforts and you would be wise to then take extra precautions when near them.

AI Autonomous Cars Must Contend With Drunk Drivers

What does this have to do with AI autonomous self-driving driverless cars?

At the Cybernetic AI Self-Driving Car Institute, we are developing AI software for self-driving cars.

A key aspect of AI self-driving cars is that they will be driving on our roadways in the midst of human driven cars too.

There are some pundits of AI self-driving cars that continually refer to a utopian world in which there are only AI self-driving cars on public roads. Currently there are about 250+ million conventional cars in the United States alone, and those cars are not going to magically disappear or become true Level 5 AI self-driving cars overnight.

Indeed, the use of human driven cars will last for many years, likely many decades, and the advent of AI self-driving cars will occur while there are still human driven cars on the roads. This is a crucial point since this means that the AI of self-driving cars needs to be able to contend with not just other AI self-driving cars, but also contend with human driven cars.

Returning to the topic of human drunk drivers, some AI self-driving car pundits have said that there is no need to contend with human drunk drivers because there shouldn’t be any human drivers allowed on the roadways.

By exclusively having only AI self-driving cars on public roads, you’d be able to eliminate the aspects of having to do deal with human drivers at all, regardless of whether those humans might be sober or drunk, since they would not be allowed to drive cars.

As mentioned, this is a rather crazy and at best naive viewpoint about the real-world.

In the real-world, we are going to have human drivers and we are going to have AI self-driving cars.

Another perspective by some pundits is that we should restrict human driving to particular lanes or roads and have AI self-driving cars also be restricted to particular lanes or roads.

The theory is that if you separate the two, meaning that you have AI self-driving cars driving on their designated roads and you have humans driving on their designated roads, you’ll avoid any kind of contention between the AI driving and the human driving.

Again, this is not a particularly practical approach.

There would be a substantive cost to set aside the lanes or roads for their appropriate designated kind of driver, whether the AI or the human, and the infrastructure costs would be relatively high to achieve this. It would also tend to imply that there are likely going to be some paths that will be a disadvantage to one or the other approach, suggesting that perhaps the AI might get roads that are going to be circuitous to get to where a passenger wants to go versus via human driven car there is a faster path (or, vice versa).

In short, I assert that we need to assume that AI self-driving cars will be mixing with human driven cars. And, those human driven cars might at times do some nutty driving, especially when a drunk driver is at the wheel of the car.

Autonomous Cars Can’t Have A Head-In-The-Sand Approach

For those pundits that are willing to concede that there will be a mixture of human drivers and AI self-driving cars on our roadways, they sometimes will say that there is no need for the AI self-driving car to do anything special about the fact that there are human drivers. In essence, they suggest that if the AI self-driving car just follows the law and properly drives, it has no need to be concerned with drunk drivers.

I call this the head-in-the-sand approach to AI self-driving car driving.

For AI self-driving cars, they should be AI-equipped to detect whether human driven cars might be driven by a drunk driver.

This involves observing how the nearby cars are behaving. I had noticed that a car was driving in the lane to the left of the slow lane, moving quite slower than prevailing traffic, and that was periodically tapping on its brakes. These are all telltale clues that the driver might be a drunk driver.

Drunk Drivers And Other Out-Of-Sorts Drivers

When I refer to a drunk driver, I don’t want you to necessarily think that the AI self-driving car will be able to know or determine that a human driver is actually in a drunken state per se.

I’m not suggesting that the AI will be able to any certainty determine the physical and mental state of the human driver that is driving a car nearby to the AI self-driving car.

The AI is merely observing the driving behavior of the human driven car. It is then a logical inference that the human driver is somehow amiss, due to the driving behavior.

Use Of V2V To Warn Other Nearby Self-Driving Cars

An AI self-driving car can be driving and not only observe potential drunken driving, but also share with other nearby AI self-driving cars what it is detecting on the roadway.

In the case of AI self-driving cars, via V2V (vehicle-to-vehicle) electronic communications, the AI of one car could communicate with the AI of other nearby cars.

In addition to communication via V2V, there is also the possibility of V2I (vehicle to infrastructure) being used for this same kind of situation. Increasingly, roadways are becoming “smart” by adding various computer capabilities and the use of V2I could allow for a roadway to communicate with the AI of self-driving cars.

The roadway might have detected that there was a driver moving at a much slower speed than prevailing traffic, and it could then send a broadcast to nearby AI self-driving cars to be on the watch for this other car.

If AI self-driving cars are not versed in detecting, predicting, and acting to avoid potential drunk driven cars, the desire to see a dramatic reduction in drunk driving car accidents might not budge much.

It has been hoped and stated that the advent of AI self-driving cars will get us toward zero fatalities, including making drunk driving accidents a thing of the past. This though assumes a world of all and only AI self-driving cars.


In a world of mixed human drivers and AI self-driving cars, we could sadly end-up with even more drunk driving injuries and deaths due to AI systems that are not adept at contending with human drunk drivers.

As such, those AI self-driving cars might get entangled into a drunk driver in a manner and frequency even higher than if there were only other human drivers on the roads.

Drunk driving is bad.

We all want to find ways to reduce or even eliminate the dangers of drunk driving. AI self-driving cars that are not able to contend with drunk drivers are going to potentially increase the risks of drunk driving incidents rather than reduce them. There will be a trade-off between potential reductions in drunk driving fatalities because people are riding in AI self-driving cars which are not being driven by drunk drivers, versus the instances of AI self-driving cars that get hit or come in contact with human drunk drivers and get into a fatality.

The AI needs to be savvy about how to detect, predict, and out maneuver those human drunk drivers.

For free podcast of this story, visit:

The podcasts are also available on Spotify, iTunes, iHeartRadio, etc.

More info about AI self-driving cars, see:

To follow Lance Eliot on Twitter:

For his blog, see:

For his Medium blog, see:

For Dr. Eliot’s books, see:

Copyright © 2019 Dr. Lance B. Eliot

Written by

Dr. Lance B. Eliot is a renowned global expert on AI, Stanford Fellow at Stanford University, was a professor at USC, headed an AI Lab, top exec at a major VC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store