Automated driving: Where will it take us?

Very rarely in life do I find myself in the scenario where I am in awe. It’s fun when it happens, it sort of reminds you of your childhood, when being in awe was a much more common experience. I remember when I was a kid, probably 10 or 11, my grandfather, who I was convinced was quite possibly the strongest person alive, told me to punch him in the stomach. It was a funny request thinking back, but a request that I can envision my 65-year-old self making to my future grandson in an effort to keep a similar narrative going.

I asked him a couple of times if he was sure, and each time, with more confidence, he assured me that I could go ahead and land a punch as hard as I wanted to right into his stomach. So, I wound up and unleashed what I thought would be an earth shattering, foundation rumbling strike. He didn’t even flinch. I was in such awe of his strength that even to this day, 20+ years later I remember the feeling.

Feelings like that don’t come around very often, but recently I found myself in awe again.

I watched the video above and was in complete awe. The background story is that the vehicle is a Tesla driving via an Automated Driving System in traffic, and as you can see, saving the driver (and other passengers?) from serious bodily injury and potentially their lives.

It wasn’t so much the visuals. I think you actually need to experience automated driving in person to really feel the type of goosebumps it will take to create lifelong memories. Rather the awe that I felt had more to do with the vision of the future. Even if automated driving doesn’t become commonplace for 20 years — it’s more likely to be much sooner, right Elon? — I’ll remember exactly when it started. In this video you can see a new future forming right before your eyes. This is awesome!

While my moment of awe will last in my memory, I also thought after watching this video about what’s going to happen in the future, legally speaking. Much like a doctor invariably walking around diagnosing people in his or her mind on the subway when he or she witnesses a cough, or spies a limp, lawyers are trained to view their world through the lens of the law.

As such, I couldn’t get out of my mind certain thoughts surrounding the inevitable scenarios which will require us to examine who will be at fault if and when an accident occurs in our new automated future, and the impact this would have on a huge swath of the practice of law.

Today, a majority of personal injury lawsuits involve claims for negligence relating to the operation of a Motor Vehicle. The simplified definition of negligence — that is to say the non-legalese version — is simply that one party failed to take proper care in doing something and through that lack of care someone, or something, may have been injured.

Care is quite the unique term. Unique insofar as it is almost primarily applicable to humans. Machines, and more accurately the software that controls their actions, are simply rule based systems that basically operate by answering — albeit at unthinkable speeds — a series of yes or no questions. These operations of “yes”, or “no” (or “1” or “0”) are the building blocks for everything we do on our computers today. I should disclaim here that I am not a software engineer or a programmer, but I have a bit of knowledge in the area thanks to a background as a technician for Apple.

These personal injury lawsuits, based on negligent actions of the driver, require that the Plaintiff’s counsel prove to a jury or a judge that the Defendant failed to exercise due care while driving in order to be successful. There are a number of ways this can be done, but more often than not there isn’t specific and direct evidence of the failure, as you may imagine. No matter what television shows choose to portray, it’s very rare to have any type of forensic, or even better video-based, evidence that will directly implicate the Defendant for acting negligently and causing an accident.

Because of that reality, lawyers spend most of their time attempting to show that the Defendant acted negligently via a litany of circumstantial evidence. Plaintiff’s counsel hopes to show in each of these instances that the accumulation of all of the circumstantial facts being produced showcases to the jury that there is no other reasonable, or “simple” explanation for the cause of the accident, other than the Defendant’s negligence. Lawyers will often rely on Occam’s Razor — the idea that given all of the possible explanations for an incident that the simplest one is the most likely — and use that in their efforts to convince a jury.

But how could that possibly work in the future with an automated vehicle? Your standard Personal Injury lawsuit, resulting from a Motor Vehicle Accident, will look a lot different for one. How does one identify the party at fault when the party is akin to an appliance, or when the failures of those appliances can be attributed to software or hardware? There are two approaches that I imagine as we move forward into this new age. First, the laws need to be updated to support the potential scenarios we may find ourselves in. Second, when an accident occurs, the strategy for litigation will also change.

Lawmakers, as they attempt to create a framework of laws surrounding automated driving, will have to focus on a breakdown of specific scenarios which may or may not occur while such automated systems are running the vehicle. Machines don’t forgo due care, software developers (or potentially hardware designers) do, but that’s not where we’ll begin. Currently and for the foreseeable future, humans will maintain an ability to override any automated driving system (herein after referred to as “ADS”) and therefore will in all likelihood continue to provide at least one explanation for a possible incident involving a motor vehicle accident in the future. Although the goal of the ADS is to remove as much human based driving as possible, human action will still represent a possibility as the case of a Motor Vehicle Accident — although not without some serious effort.

Lower level ADS can be seen in the form of automatic braking, adaptive cruise control and several other forms of new “automatic” safety mechanisms which attempt to inform a drive of an unsafe situation, and then to take over when the drive fails to respond appropriately. Even if a human being, in their slight time behind the wheel is able to, surprisingly, overcome all of the baked in safety mechanisms meant to avoid a Motor Vehicle Accident in the future, given the new hardware and software systems in place, ADS may make proving such negligent action easier than ever before.

It wouldn’t be hard to imagine companies who provide the software, or the hardware, or the vehicles for that matter, establishing a constant stream of data, saved to the cloud of course, that they or others (via discovery) could utilize obtain in an attempt to gain a contemporaneous viewpoint of the operations of the vehicle at the time of any accident. This will appear very similar to an upgraded “blackbox” that planes used today and may be a requirement in the future — imposed by state or federal authorities — in an effort to aid in the investigation of any failures of an ADS. This could be a huge boon for parties (for one at least) as they attempt to resolve any claims of negligence involving ADS as to whether or not they may have involved human or non-human entities. This would also help to streamline and potentially revolutionize Discovery in these cases as it specifically relates to potential liability. This beneficial offshoot of ADS may shed light onto other potentially revolutionary aspects of the automation movement we’re now beginning to see.

Alternatively, if it can be shown that an accident occurred without any type of human interaction, the software running the automated systems may be operating in one of two states. First a compromised state, where a software bug has rendered the system incapable of executing all of its necessary functionality to maintain a safe driving experience, as designed. This is hard to imagine only because you would have to think that a pre-check occurs within any ADS to verify both it’s hardware & software at the time of departure, leaving spontaneous failure or environmental reasons as more viable alternatives. It is also likely that redundant systems are in place to “pass off” or return driving to a human should this failure occur during transit or even if a human in unavailable, additional software redundancies to safely (if possible) take the car and place it out of harm’s way given the surroundings. I won’t be getting into the ethics debate in this post about what the vehicle should do if it can’t do something safely.

These redundancies are also likely on top of an additional set of hardware redundancies built in to help ensure that anything less than catastrophic system-wide hardware failure would be sustainable for a varying amount of miles. But this is hard because these failures are unpredictable, erratic and with the above-mentioned built in systems for redundancy (hopefully) built into vehicles, a spontaneous error would almost be unfathomable, but not necessarily impossible. Environmental issues, i.e., another vehicle causing an accident or a weather condition, causing an accident is much more plausible and may test the boundaries of neural networks and algorithms as they attempt to learn and gain experience on all possible road conditions or inadvertent impacts caused by forces unrelated to the vehicle. I wonder what the limits are for safety actions that an ADS may take — but again another topic for a another post.

Secondly, the software could be in a non-compromised state, with the software itself is running as intended — however an incident has still occurred in light of the fact that the design of the software was faulty. This would be a much more plausible scenario. Humans will also provide a source of error in the chain of custody surrounding the proficiency of software and an ADS is unlikely to be different. Alpha and Beta testing would undoubtedly be stringent as it would relate to an automated system, but perfection is almost assuredly unobtainable and this scenario presents a conventional type of negligence question, as previously discussed. Did the engineer or Team Lead act with due care? What would happen if a major automotive manufacturer was found guilty of negligence in light of their ADS, especially when the reliance of for the majority of human transportation shifts to software? (I do wonder here if an open source movement would provide some respite)

I’m not worried about a high prevalence of imperfections in the systems themselves, but it’s a simple fact of life with software — nothing is ever perfect. Really, it’s more of a truism with life itself in my experience. This less than perfect won’t be the difference between a dead battery at lunchtime, but may instead lead to a crash or injury. I lack all of the necessary information & expertise to draft up a treatise on how to avoid this scenario, but working together, companies pursuing ADS and regulators should be able to provide software that still produces a driving experience that is orders of magnitude safer than what exists today.

These scenarios present three identifiable states of error (human, compromised and non-compromised), which may lead to an Motor Vehicle Accident involving an ADS. These scenarios can be investigated, broken down and regulated accordingly by state and local authorities, much in the same way that Negligence is today. But once these new frameworks have been identified and managed, what will happen to the Plaintiffs who are so prevalent in today’s Personal Injury Civil suits?

One of the reasons that attorneys are so well acclimated to litigation involving motor vehicle accidents — and the reason it immediately came to mind for me — is of the sheer numbers of Motor Vehicle Accidents that occur, which then progress into litigation. The United States experiences around 1.9 million motor vehicle accidents per year where injuries occur. The realistic goal of the automated driving future is to produce a major reduction in this number, and not to guarantee that no accidents ever occur again — although this is what in part drives the endeavour to create ADS. The real metric to determine success is observe the risk involved in the use of a motor vehicle being reduced to such a level that accidents occur with such rarity that they themselves are seen as unfamiliar and tragic — similar to how we view accidents with major commercial airlines.

Many attorneys around the country focus part of, if not a majority of, their practices around litigation involving Motor Vehicle Accidents. In a future where automated driving reduces the number of accidents, by some estimates up to 90%, these cases would become so infrequent that many attorneys will see a seismic shift in their practice patterns in the future. If the market for such cases was to vanish — as it appears it will — how would Attorneys handle it? (Future changes in the market for Attorneys will have to be reserved for another blog post, another time). With less demand for such services, it can safely be assumed that this area of practice will become stagnant, forcing many Attorneys to focus their attention elsewhere.

Even more adaptation will be necessary with the transition from a primary focus shift from human error to software or hardware malfunction. These types of lawsuits will quickly transform into Product liability claims, a speciality of its own. But even more drastic would be the scope and size difference between the Motor Vehicle Accident Civil Cause of Actions today and those of the future.

When your local attorneys shift their focus, boutique and large firms will take over their role as the primary litigators for most, if not all, of Motor Vehicle Accident litigation, and will need to ensure such cases are worth their time. Class Action Suits feature numerous Plaintiffs and normally, given that fact, a much higher level of complexity, along with an elongated period of recovery for Plaintiffs and in lower amounts. The closest viable corollary case today may be that of Takata Corp, and the several automobile manufacturers who faced product liability claims over faulty airbags in 2015.

The complexity aspect is particularly consequential in the automated driving realm. Historically, it’s been relatively easy to explain how negligence exists during a motor vehicle accident, in light of the common experience of driving. People inherently understand that when you see a stop sign or a red light you come to a stop. People also understand how a four-way intersection works and that going too fast around the corner can easily lead to the loss of control. The shared experience of driving allows jurors to easily understand what negligence may look like in a standard motor vehicle accident case.

Expanded further, jurors can normally comprehend how a defect in manufacturing has come about through the careful and thoughtful testimony of an expert who helps explain the problem in either the design or the manufacture of a specific component, how it came to exist and the impact that it has. Jurors normally find such information consumable in part because they can fathom what it is that the experts are discussing or the systems surrounding the parts are readily absorbed by an average person. But, if you’ve ever provided your family impromptu IT support during the Holidays, or hung out at the Genius Bar for a bit, it’s easy to see that the complexity of software and the associated components (networking, cloud storage, etc), even as ubiquitous as they are in our everyday lives, are elements of a series of systems that many people actively avoid competency in. Even today, watching Attorneys, Judges, and non-expert witnesses attempt to understand even current consumer levels of technology can be comical.

But beyond the comprehension of the sophisticated systems in place, these class action lawsuits will have varying degrees of damages. As personal injury attorneys will readily acknowledge, after the question of liability, next comes proving your client’s damages. While investigating the causes of the crash may fall under the framework hypothesized above, the nature of the damages will certainly open the possibility of very serious cases where different Plaintiffs have suffered in wildy varying degrees. But if they transition into a standard class action style, will clients even be interested in partaking given changes mentioned above? Will they recover reasonable damages for their specific Personal Injury?

Even outside of the case mentioned above regarding the defective airbags, technology companies are routinely sued in products liability cases which may provide an accurate depiction of where Motor Vehicle Accident litigation may be heading. A common headline strewn across the internet today is “Apple hit with class-action lawsuit for ‘Issue A’.” Google, Microsoft and many other technology companies have weathered similar attacks when unhappy customers encounter an issue with their software or device. Will the lack of accident based litigation be replaced with feature centric suits from discontented users?(Another topic for another time, how many of these lawsuits will come about when someone doesn’t like a particular route or speed their ADS utilizes?) Even Tesla has felt the sting of this now almost standard practice.

But rarely do issues involving software bugs, storage space, Wi-Fi or any other component or attribute to software involve matters of, potentially, serious bodily injury or even death. Suffice it to say that these issues deserve serious discussion.

The point here isn’t to say that the automated driving vehicle will be more likely to injure an occupant, in-fact just the opposite. With ADS themselves working to reduce Motor Vehicle Accidents to 10% of their current amounts and under the assumption of continued greater refinement of those systems utilizing current AI advancement techniques, alongside continued safety innovation in materials and automotive design keeping it’s current pace, in the not too distant future driving with an ADS may be the safest form of transportation available.

But the possibility for the potential of serious bodily injury or death has to be discussed. In the unlikely event such an incident were to occur, how would the public react if and when they heard that a software bug’s worst attribute went from potentially missing your morning alarm to injury or even death? How can we ensure that victims in an accident will have access to the data and adjudicative measures necessary to enforce their legal rights?

I’m not sure what’s going to happen in the future. But I have a lot of questions. How will income inequality affect the proliferation of the self driving vehicle? Will the wealthy be the safest drivers for 10 or 20 years until the cost of this new technology comes down? Or is this the type of revolution that everyone will use almost immediately?

I think that the ubiquitous personal injury lawsuit stemming from a motor vehicle accident will be a thing of the past. This revolution will spark a massive change in the legal industry that may help re-shape it in a manner unseen for sometime. Greater availablity of data steming from constant data streams flowing from automated driving systems will help revolutionize these cases (and others — especially criminal cases in my estimation) in the rare event that an actual accident occurs. This revolution in the accumulation of data may be simply beginning, another impending change brought on by the proliferation of AI in our world (Think of all of the data that the Internet of Things will inherently create).

I can also see the suppliers of the software, hardware or the vehicles themselves becoming frequent targets for class-action suits, especially as they release and innovate on their products, whether an actual issue exists or not. The real question is will these types of suits, with all of their already alluded to issues, truly provide the type of recourse that will allow victims in these rare accidents to be fairly compensated? That is the question for the future. Will regulators and legislatures find a framework to handle this new, awe-inspiring reality, and given the historic reduction in accidents, will it even matter?