Self-Driving Cars Will Fail

by: Austin Craig


Self-driving car technology is not yet ready for prime time. Driver assist is.

The Legal challenges and potential liability are immense.

How to play the hype.

Currently, self-driving cars are the rage in technology circles. We do believe that, in certain environments, they could work well (such as closed-loop environments, i.e., a mining operation or a shuttle service on a closed track at an airport). But like other fads (Google glasses, nano tech, 3D TVs), we think self-driving cars will take more time to mature than many realize due to their inherent legal issues and safety concerns.

That's not to say self-driving cars can't happen. Perhaps eventually. However, we think it will be quite some time before they are "smart" enough to be on the roads safely and we are not alone in our assessment.

"You can't just assume this stuff is going to work," says Phillip Koopman, a computer scientist at Carnegie Mellon University who works in the automotive industry.

However, while we might have serious safety-related reservations -- this does not mean auto makers share our view. Clearly the "Me too train" has left the station and many car manufactures including Ford (NYSE:F) and Tesla (NASDAQ:TSLA) are getting on board.

Safe you say? Take a moment and come up with your own custom ridiculous situation and think, "Would I want to let my car try to make the right choice? Would I trust my life and my family's to such devices?" Note: We do think that driver assist features such as correcting back onto the road will catch on and make sense. More on this later.

Legal Challenges

Right now various tech companies such as Intel (NASDAQ:INTC), Nvidia (NASDAQ:NVDA) and AMD (NASDAQ:AMD) are all pushing various forms of deep learning or are fighting for a role in self-driving vehicles. Yet, for all the pushing of this new-fangled technology, has anyone really stopped to think of the enormous liabilities they could be held for?

nvidia deep learning car

Assume the car has to make a decision between the survival of the occupant or the approaching car. Which does the car pick? If the car picks the occupant then the other parties' survivors have legal grounds to sue. If the car picks the other party, then the occupants of the car have grounds for a lawsuit. It is a no-win situation for the car company and the software/hardware developers. Now, we can already hear Silicon Valley preaching "It's deep learning! We can teach it!" and "It has cameras and radar!". Well, indeed, but can they teach it to react (or not react) to -any- conceivable situation?

The answer is no. Deep-learning cannot come up with a solution to every single possible situation, as they are infinite. The car has to "learn" through data, trial and error, programmer feedback, and simulations on how to react. This is simply an impossibility for it to make the correct decision to any situation that might arise.

The fact is that they will wreck and people will be injured, if not worse. The potential liabilities are huge. Perhaps in the end it will come down to legal liability vs profit potential. Assuming the mentality of an insurance company, it may simply be cheaper to just push forward and pay out claims as accidents happen. However, when claims exceed profits...

Human Nature

Additionally, people will be people. Certain auto manufacturers (such as Tesla) say the driver needs to be alert, hands on wheel at all times, and ready to correct the vehicle when they make mistakes.

(Tesla Self Driving Car)tesla self driving car

That is not human nature. People will sleep, surf the internet on phones and tablets, and do just about anything but sit in a chair doing nothing as the landscape rolls by. Some are even absent from the driver's seat. Watch videos here. To top that, if you call your car "autopilot" (as Tesla does), it sets up certain expectations with consumers. People will treat and use the vehicle as what the name implies.

People are not going to buy an "autopilot" car and treat it like a driver assist car with the hands on the wheel all the while alert for danger. Honestly, the only way any of this robotic car argument makes any sense would be if you are driving late at night, fall asleep, and start to run off the road. The car is programmed to correct this situation for you and surely that is a good thing. But as far as self-driving cars-- it is an unsafe, pipe dream. Sure, we can put the cars on the road, and yes, they are pretty okay at driving on an interstate or simplistic courses like Nvidia's car did here but when the legal liability catches up with the companies, expect massive payouts of capital to damaged parties.

Last, as per the Observer, is it possible self-driving cars can be programmed to break the law in order to avoid accidents? Would it have responded correctly by not stopping and potentially causing an accident at 1:31 into this video if a car were behind it? Could it have broken the law and gone into the opposing lane to avoid the people? Will those algorithms be programmed to solve ethical questions? Can it decide on which car to avoid when given the situation hitting 1 of the 2 cars is unavoidable? (make up your own ethical situation and it gets sticky)

Winners and Losers

With every fad, you have winners and losers. The losers here are the car companies. Pouring hundreds of millions into a product that is legally-challenged from the start is not wise. However, because they fear being left behind, they are tossing greenbacks into this money pit. Avoid the car makers.

Currently, the following have announced various self-driving car plans or projects. Click the name of the company to see the link. Honda (NYSE:HMC), Toyota Motor Corp (NYSE:TM), General Motors (NYSE:GM), Fiat Chrysler Automobiles (NYSE:FCAU), and lastly Ford. Of course not everyone is drinking the self-driving Kool Aid. Ferrari (OTC:RACE) has publicly stated they will not jump on the fad train.

"There will be no autonomous Ferrari in the foreseeable future." and "When you are driving a Ferrari, we want you to concentrate."


Moving on, do car companies actually think they will sell more cars? It's not like hoards of people are on the sidelines in horse-drawn buggies and carts holding out for self-driving cars. No, if a person wants a car, they own a car currently. This is simply a technological cold war a.k.a. my neighbor car company is working on it, so I better get on it as well.


The winners will be the technology companies that have deep experience, intellectual property, and existing technology that can be leveraged into cars. AMD, Intel, and Nvidia all provide graphic cards that can be utilized or built upon. These are the companies that can ride out the fad for as long as the auto makers will play the game (And wow are the auto makers ever playing the self-driving car game! They hype levels are at least equal to 3D printers from a year or two back.)

nvidia ceo

(Nvidia presenting Xavier - The brains of Nvidia efforts in cars)

Of course, the above 3 companies will be included in the lawsuits under the theory of product liability as well. A lawyer once explained that theory as such: The idea of what she called "spaghetti" -- you sue everyone even remotely associated and leave it up to the judge to determine what sticks when you throw it against the wall of the court. In other words you sue everyone to see what defendants are brought into the case and which ones are let off the hook. (Lawyers on SA feel free to add color to this.)

Having provided the "brains" of the the self-driving cars, you can bet Nvidia, Intel, and AMD will be involved in many legal battles once self-driving cars hit the road (and other objects).

However, since these three companies are leveraging existing gaming graphic card technology, the research and development cost is inexpensive as opposed to starting from scratch. Yes, it does cost millions upon millions, but if you are selling hundreds of millions in sales like Nvidia is to the automotive sector, then benefits exceeds cost. Long term though, it might be driver-assist that takes over and thus provides legal relief for all of the companies once everyone realizes that completely self-driving cars can never be completely safe. Yet driver-assist might actually improve safety by gently correcting a sleepy driver back into the lane at night by taking control of the vehicle to point it back onto the road.


If we are dead set on joining the self-driving craze, we need to invest in AMD, Intel, or Nvida. These three can leverage existing technology in order to tap this craze. Long term, driver-assist will catch on and does make sense. Self-driving cars do not make sense from a safely and liability standpoint.

The hype train will continue to pump these stocks and we can profit from that until the public realizes that it's an unsafe dream.

However, all three companies have rock-solid products. The auto manufacturers are to be avoided. It's not like they will not all have roughly the same technology in a few years and the car market is saturated as it is. The takeaway here, avoid the car makers, buy the stocks that provide the brains that go into the car for driver assist growth (AMD, INTC, NVDA) but realize that fully self-driving cars are a legal and safety nightmare.

Additional Links

For a wonderful cringeworthy time visit this moral simulator.

It is full of delightful moral situations where no choice is good and it provides a recap in the end of how you compare to other people.

Disclosure: I am/we are long AMD.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

About this article:

Author payment: $35 + $0.01/page view. Authors of PRO articles receive a minimum guaranteed payment of $150-500.
Want to share your opinion on this article? Add a comment.
Disagree with this article? .
To report a factual error in this article, click here