Non-Stop AI Self-Driving Cars: Truths and Consequences


By Lance Eliot, the AI Trends Insider

How often do you use your car?

I am betting you likely use your car in the morning to drive to work, and it then sits there, parked, until you drive home after work. For much of the time, you aren’t actively using your car and instead it sits, waiting at your beck and call, and like a faithful steed is ready to go whenever you are (well, assuming it starts, and that it is well maintained). When you think about this, you might realize that you’ve purchased an asset that most of the time is unused. It’s a resource that remains idle most of the day and night. It also loses value over time, decaying or otherwise aging, and becomes less usable and costlier to maintain.

Given that there are about 200 million cars in the United States alone, one could make the argument that we are a seemingly selfish society in that we each assume or demand that we need to possess our own car, which uses up the collective riches of the society, and yet this costly resource is only utilized a fraction of the available time. For most people, their car is their next costliest purchase, second to the purchase of a house.  Thus, this is a big economic aspect and it’s not like we are talking about having your shoes idle for a lot of the time that they could be used. Cars are expensive. It’s a big deal.

Estimates are that our cars are parked 95% of the available time that they could be used (using 24×7 as the “available time” metric). This staggering 95% means that we are only using our cars about 5% of the available time.

You might argue with me and assert that you wouldn’t be using your car when you are asleep, so for me to suggest that your car could be used 24 hours per day is maybe unfair. Okay, let’s subtract 8 hours out of a 24-hour day and say that we’ll grant that cars won’t be used while the owner is asleep (we’ll revisit this assumption later on). In that case, cars are not being used an estimated 92.5% of the time versus 95%, and thus being used only about 7.5% of the time (this is based on surveys that suggest we use our cars for 1.2 hours per day, and so for a 24-8 =16 hour day it would be 1.2/16 = 7.5%).

Multiply this by 200 million cars and you have a staggering sense of the magnitude of the resource that is idle most of the time. Is this a misuse of these resources? Are we denying ourselves a greater economic value for the steep investment we put into these resources? There appears to be untapped value that we could leverage to the benefit of us all, both collectively and on an individual basis.

How can we solve this problem? Generally, the biggest constraint that holds back the more effective use of the resource is one thing, the driver.

A car does you little good if there isn’t a driver. Without a driver, it’s a parked car. It’s a paperweight. Sure, you can store your stuff in it, and you can even use it as an in-car delivery mechanism. You can stare at it. You can admire it. You can show it off to your friends. But, other than serving in this limited ornamental fashion, it has little it can do when there isn’t a driver.

Now that I’ve stated the problem in those terms, we can begin to find ways to solve it. If the constraint is the lack of a driver, perhaps we need to make available more drivers, or find more drivers that could use the resource. This though creates other problems. If I let my friend, Joe, use my car when I am not using it, will my car be available when I really need it? The great thing about owning a car is that you can usually have it at your beck and call. Once you’ve parked it, you know it will be there, unless maybe you are illegally parked and it gets towed away.

We want then to retain the benefit of the car being immediately available, when needed, if needed, and control its individual availability. At the same time, we need a driver in order to make it more effective as a resource. Just posting on Facebook that my car is available after I get home from work and until I need it in the morning, and offering it to anyone else, well, yes, there are specialized sites that provide this social networking, but it’s a pretty small amount of people that do this. As mentioned too, if I let another driver have my car, there’s not much I can do to make sure that they will bring back my car the moment I need it (suppose I wake-up at 2 a.m. and just absolutely need to drive over to a burger place and get myself a late night snack, while Joe has driven my car to the other side of town for bar hopping).

Imagine if you could have a driver always available. A willing chauffeur that sits in your car, at all times, and can be invoked by you, as needed, when needed. Could you hire someone to be this chauffer? You could, but it would likely be costly (you’d need to pay them, and maybe cover their expenses, and if it is just one person they might not like sitting in your car for 24 hours per day, each day, waiting to drive you – those pesky humans!). Let’s instead shift our focus from a human chauffer to instead using automation that acts like a chauffeur.

Would you still use your car for only 5% of the time that it could be used?

There are some people that might not change how they use their car. They might use their car in exactly the same way they do today. On the other hand, wouldn’t it be tempting while at work to have your car go get you some lunch, so that you could just come down from your office and meet up with your car in the parking lot, and it already went over to get you pizza. Or, maybe your kids need to be picked-up from school, and instead of having to rush your guts to get to them, you calmly send your car to pick them up and then take them home, and then the car comes back to your workplace.

That’s the promise of AI self-driving cars.

To clarify, that’s the promise of true AI self-driving cars. A true self-driving car is considered a Level 5, which is a self-driving car that can drive in whatever manner a human can drive a car, and for which there is no human intervention needed to drive the car. For cars at a level less than a 5, a human driver is still needed in the car. Therefore, when referring herein to a self-driving car that has its own internal chauffeur, please keep in mind that I’m alluding to just the Level 5 true self-driving cars.

If you read the major news outlets coverage of self-driving cars, you might think that just any kind of self-driving car is going to get us to this potential non-stop 24×7 capability. Not really. It’s the true Level 5 self-driving cars that will do so.

Furthermore, the major news outlets tend to gush at the idea that our cars can be used 24×7. Yes, it definitely will allow us to better maximize this expensive resource. It will tap into that untapped value. The benefits for society are impressive.

Wear and Tear on 24×7 Self-Driving Car a New Challenge

One question that does not seem to be getting any attention involves the aspect of whether our cars can cope with being run non-stop 24×7.

Think about the amount of wear and tear that your car gets today. Whatever amount it gets, this is based on 5% of usage. Will that car be able to cope with adding an additional 95% of the usage time? How long will your tires last? How long will it be before you need to change oil on the car? How long will it be before some of the engine parts break down? And so on.

Our present day cars are not made to work in this kind of non-stop 24×7 mode. You might assume that they can, but this has not yet been tried on a massive scale. Suppose we switch over to all AI self-driving cars, somehow, magically, and thus we take the 200 million conventional cars that are now being used only 5% of the time, and we wave a magic wand and we now have 200 million AI self-driving cars that are running all day and night, day after day, for 100% of the time.

Yikes, I’d bet that we’re going to have those AI self-driving cars visiting the maintenance shop, very frequently.

You would also assume that the degree of breakdowns is probably going to increase. In other words, if today you only have your carburetor go out every so many years, it’s going to happen every so many weeks or months instead, and possibly cause other parts to also wear out or break. You can’t just assume that the car failures of today will simply be more frequent, and you likely need to consider that they will be more frequent and of a greater depth and degree.

Today’s cars are not designed and built with the expectation they will run non-stop 24×7. We don’t even really know how cars will react to this kind of bruising usage. Car parts aren’t built to take this kind of punishment. We must then consider that either we accept the idea that our AI self-driving cars are going to have lots of physical failures and failings, simply due to the extended use, or we need to have the auto makers take another look at the makings of a car and figure out if somehow cars need to be redesigned and differently manufactured to cope with the non-stop 24×7.

If we end-up having to completely redesign the physical aspects of a car, in order to accommodate the desired non-stop 24×7, the odds are that the cost of an AI self-driving car is going to go up. Rather than simply using a conventional car that has the added AI self-driving car elements, we might need entirely new cars, ones that are built from the ground up to withstand such high usage.

I’ve heard some AI self-driving car pundits say that even if the maintenance needs rise, so what, since you can just tell the AI self-driving car to go to the maintenance shop to get repaired. Maybe we’ll have maintenance shops that work late at night, and our AI self-driving cars will all drive to those maintenance places while we sleep. Notably, we’re once again cutting out some of the available time to use the AI self-driving car, and so we need to keep in mind that we are reducing the use of the resource. It’s not going to be non-stop 24×7 anymore, and maybe more like 20×7 if we assume that on the average it needs to be in the shop some number of hours per day (this does not suggest it will happen each day per se, but that it might be in the shop for an entire day and so we are apportioning that over the course of a week).

Let’s assume that we will go along with the pundits and agree that having the AI self-driving car go to the shop, on its own, relieves the human owner of the burden of dealing with the maintenance. What about the cost of this maintenance? You can imagine that if your AI self-driving car is continually going to the shop, the costs of owning and maintaining the car is going to go up, quite a bit.

Not only is the cost of maintenance going to rise, presumably a lot, we can make the assumption that the number of miles put on the car is going to go up, exponentially. This also tends to imply that the value of the car will drop, fast, as it rapidly becomes a more used car. Today, we all realize that the moment you drive a car off the lot after buying it that the value immediately drops. Once you’ve got your non-stop 24×7 AI self-driving car going for a few months, imagine how much value it will have lost.

We also need to consider the fuel costs. If the AI self-driving car uses gasoline, this non-stop 24×7 use will have a voracious appetite for gasoline. Your fuel costs are going to go through the roof. What impact will this have on the price of gasoline? If the demand for gasoline rises due to the AI self-driving cars on the road, is there even sufficient supply available to meet the demand. Will the price of gasoline go up, since it will be in such high demand?

I know that some of you will say that we should be getting rid of our reliance on gasoline and that AI self-driving cars should be electric instead. Generally, yes, we’ll likely be seeing that AI self-driving cars are going to go the electric route. You then need to add the cost of electricity for charging these cars. And, the length of time to charge a car will need to be factored into this, thus again reducing the amount of available usage time of the AI self-driving car.

Hidden Bugs Could Surface with 24×7 Self-Driving Car Use

Another interesting aspect of the non-stop 24×7 approach involves the potential for encountering bugs or errors in the AI systems of the AI self-driving car. Allow me to explain.

You might be assuming that the AI self-driving car is going to work perfectly, other than those annoying physical breakdowns that might occur such as the carburetor or the brakes going bad. I hope you are sitting down when I mention that the AI systems driving the car are not going to be 100% perfect. There is a chance that there will be errors or bugs in the AI system.

If the AI self-driving car wasn’t driving around much, the errors or bugs might not either be encountered or rarely encountered. Once the AI self-driving cars are going non-stop 24×7, the odds rise that at some point a hidden bug or error will be invoked. This means that the AI self-driving car might do something wacky, or untoward, or have other ill-suited behavior. The consequences can range from nominal to severely adverse (including crashing or getting into an accident).

By-and-large, the AI self-driving cars will have Over-The-Air (OTA) capabilities, meaning that bug fixes can be pushed down into the AI self-driving car. This though is somewhat misleading because first we need to know that the bug or error exists, and when it first arises it could happen unexpectedly and seemingly randomly until traced to its origins. There’s the time needed for the auto maker or tech firm to craft a fix. There’s the downtime of the AI self-driving car as it is waiting to get the fix (if its safest to stop the usage until a fix is available), and the downtime when it actually gets the OTA update. There’s also the chances that once it gets the OTA update, it might intermingle with some other aspect of the car that creates other new issues.

With the non-stop 24×7 AI self-driving car being on the road so much, we can expect that breakdowns while on the road will rise. Your AI self-driving car is going through the drive-thru at McDonald’s for you and was intended to head back to your home to deliver the meal to you. Unfortunately, due to the non-stop usage that’s been happening for months on end, all of a sudden your AI self-driving car breaks down at the McDonald’s window. You need to get a tow of your AI self-driving car to a maintenance facility.

Let’s therefore add the cost of towing to our basket of costs for the non-stop AI self-driving car. Plus, since we are saying that the AI self-driving car will be going 24×7, your AI self-driving car might breakdown at 3 a.m. and need a tow at that time. This could be expensive to get a tow at oddball times of the day or night. You might say, well, just let the AI self-driving car sit wherever it is until the normal tow time arrives, but this assumes that your AI self-driving car has not broken down in the middle of a road and blocking traffic, or that you otherwise need to get it back into service as soon as you can.

I’ll further up the ante by pointing out that the non-stop 24×7 AI self-driving car is going to get into accidents more frequently due to the 24×7 usage. There are some AI self-driving car pundits that insist we’re going to have “zero fatalities” once we have AI self-driving cars, implying there won’t be any accidents at all. Nonsense. First, we are going to have a mixture of human driven cars and AI self-driving cars for the foreseeable future.  Therefore, we will have accidents involving both human driven cars and AI self-driving cars. It is inevitable. Second, an AI self-driving car is not a tank, in the sense that if some kid tosses a rock off an overpass and it hits an AI self-driving car, the AI self-driving car is going to suffer damage.

The rise in accidents involving AI self-driving cars that are non-stop 24×7 will increase the need for the AI self-driving cars to be in the maintenance and repair shop, and thus increase the amount of downtime of the AI self-driving car. Once again, it will raise the costs of owning an AI self-driving car.

Take a macroscopic view and consider what these non-stop 24×7 AI self-driving cars will do to our transportation infrastructure. Our roads are barely tolerable today in terms of potholes and other roadway surface and structure issues. With say 200 million AI self-driving cars driving all over the place, non-stop, 24×7, you can bet it will undermine the roads and we as a society will need to spend more money on our transportation infrastructure. You can take that cost and apportion it to all of the car owners and impute that their ownership has a higher cost due to covering our roadway system for this increased usage.

By now, I hope you are convinced that the idyllic world of non-stop 24×7 AI self-driving cars is filled with lots of rubber-meets-the-road pitfalls and added costs. There isn’t such a thing as a free lunch.

Speaking of costs, we don’t yet know what the cost of a true AI self-driving car is going to be. Let’s assume it is going to at least at the start be costlier than conventional cars. The question then arises as to who can afford these fancy, expensive AI self-driving cars. Besides the purchase cost, let’s pile on top the added costs due to the increased maintenance and repairs, the added fuel costs, and so on. The affordability would seem to exceed that of the average person to buy such a car. Will then AI self-driving cars become a rich person’s vehicle and not be available for the masses. Will these create ill-will and a sense of elitism about AI self-driving cars.

There’s another way around this, namely that people will buy an AI self-driving car for both their own needs and to have a ridesharing revenue generating resource (often referred to as Metered Access to Shared Cars or MASC).

This means that a potential buyer of an AI self-driving car might be able to get a loan to cover the cost of the AI self-driving car, since it will essentially pay for itself over time. This raises other issues such as changes in accounting rules for how we treat these revenue generating cars for tax purposes and as a capital investment. The average person might end-up forming a company to help purchase their revenue generating AI self-driving car, rather than making the purchase as an individual. There are open questions about how the ridesharing will be arranged, such as via using a conventional service like an Uber or Lyft, or maybe via social media such as Facebook or Twitter. Time will tell.

Providing your AI self-driving car as a ridesharing capability will aid in the affordability factor, and it will be a means to fill-up the 24×7 of intended non-stop usage. While you are at work, you rideshare out your AI self-driving car. While you are sleeping, you rideshare out your AI self-driving car. This though does begin to put a pinch on your desire to have your AI self-driving car available at your beck and call. Suppose you’ve sent it along to do a ride for someone headed to the airport, and meanwhile you suddenly have your spleen burst and want to have your AI self-driving car urgently drive you over to the local emergency room. It will be somewhat tricky to consider how much of the time you want your AI self-driving car doing ridesharing versus being parked and not earning income, but ready immediately for your use.

There are additional ways to also derive revenue from your AI self-driving car. Some believe that AI self-driving car will be outfitted with LED’s and displays in them, so that you can watch movies and other entertainment while in the car.  It seems logical to assume that we’ll all be spending lots of time in our AI self-driving cars and doing so with no need to pay attention to the road, thus we’ll want to be entertained. Some even think that we’ll want to make classrooms inside of our AI self-driving cars, using our in-car time by taking online classes and earning college credits or getting certifications, doing so a few hours at a time on our daily commutes to work.

This in-car extravaganza of online displays provides an opportunity for having ads shown to us. Therefore, besides renting out your AI self-driving car via ridesharing, you might also be willing to have ad placement for whatever is being shown inside your AI self-driving car to occupants that are using the AI self-driving car. We’ll have to see whether occupants are willing to tolerate this. The good news is that it might mean that the ridesharing cost to the occupants possibly goes down, as long as they are willing to endure tons of ads. The ad revenue comes to you, the owner of that expensive AI self-driving car and can offset all of those ongoing costs for keeping the machine running.

The emergence of true Level 5 self-driving cars will undoubtedly convince us to use our AI self-driving cars on a non-stop 24×7 basis. It sounds logical. It seems smart. Unlock the hidden potential that exists today in conventional cars and remains untapped. As mentioned, the benefits of having a 24×7 advanced car seem relatively apparent, while the added costs and downsides are rarely pointed out. I am not casting aspersions on the notion of the non-stop 24×7 AI self-driving car. Just trying to make sure we all have a realistic understanding of what’s to come next. And, with that realization, start planning to ensure that we can really achieve the non-stop quest and not get stopped along the way there.

Copyright 2018 Dr. Lance Eliot

This content is originally posted on AI Trends.