By Lance Eliot, the AI Trends Insider
Standards are important. Yet there are always qualms about whether a particular standard is up to the task at hand. Perhaps a standard is obstructing further progress or becoming haggard and no longer especially productive.
Indeed, a valid argument can be made that sometimes a standard gets in the way and has to be revised or entirely revamped. Out with the old, in the with the new, as they say. A standard might have outlived its usefulness or at least need some hefty repainting and well-needed shop repairs for it to continue forthrightly as a guiding light of sorts.
One particular standard has been attracting attention lately, though you might not be aware of the slowly rising din unless you are closely watching the tea leaves. Yes, bubbling in the inner circles are some grumbling insiders and outspoken pundits that are starting to overtly rankle about whether the time has come to cut off the topmost level known as Level 5 that constitutes the pinnacle of the autonomous levels for self-driving cars.
Yes, an increasing undercurrent is pushing to get rid of the Level 5 part of the scale. Cast aside that darned thing, they are clamoring. Get rid of it. No replacement is needed and nor desired. The new topmost pinnacle would become the existing Level 4. Thus, whereas today we have six levels of autonomy, scored from zero to five, the total number would drop to just five levels and the shinning apple at the top would become Level 4.
I realize that some of you might not know what in the heck this is all about.
Let’s back up and make sure we are all on the same page about the nature and scope of the prevailing standard regarding the levels of autonomy for ground-based surface vehicles (i.e., self-driving cars). Once we’ve gotten all of you up-to-speed, we can then dig into the gripes, complaints, and at times heartfelt qualms about that revered Level 5 and the contention that we need to heave-ho with it.
In my intrepid efforts as an ongoing self-driving cars whisperer, you might be aware of my numerous in-depth analyses of the existing and firmly implanted standard concerning the levels of autonomy for driverless cars, also referred to as self-driving cars or at times simply as robo-taxis. For my closeup assessment of the latest and recently released version of the venerated standard, see my column coverage.
In brief, some adamantly assert that the Rosetta Stone for defining the nature of self-driving capabilities is the esteemed SAE International standard formally as the “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles.” Those that are in the know refer to the standard by its official designation, namely J3016.
Why is this standard so important?
You see, there is little doubt that the future of cars consists of AI-based true self-driving cars. No human driver is involved in a true self-driving car. Keep in mind that true self-driving cars are driven via an AI driving system. No need for a human driver at the wheel, nor is there necessarily a provision for a human to drive the vehicle.
It sure would be nice to have a means to across the board make use of an established scale or rating system that would enable us to all collegially gauge the degree of autonomy that a vehicle has achieved.
Thus, this is a key genesis for the crafting and subsequent promulgation of the SAE J3016 standard.
By having a codified standard, we can all talk the same lingo, as it were, and be relatively assured that something is substantive and tangible about the terminology we are using. Without such a standard, we would all be likely talking past each other and using our own idiosyncratic vocabulary that could mean one thing to some and mean something entirely different to others.
By having a standard that delineates the levels of autonomy, we can all readily discuss and discern the status of self-driving car efforts.
In theory, by using the provided scale, we can separate the wheat from the chaff.
That being said, please be aware that some vendors have blurred the lines by using questionable naming for their wares and conspicuously avoid citing the scale to apparently avoid getting pinned down on what their vehicles actually achieve. See my columns for indications in which I take to task those purveying such unsavory efforts.
Added to that devious morass is the underhanded ploy of using the scale by ascribing an existing vehicular capability to what the vendor hopes their vehicle will someday become, rather than getting jammed up with the otherwise demoting connotation of what it can actually do today.
Okay, that’s a bit of helpful grounding on this topic.
Here’s an intriguing question that is worth pondering: Why are some carping about the standard for autonomous levels and seeking to jettison the topmost Level 5 from the widely known self-driving cars scale?
First, I’d like to quickly further clarify what is meant when I refer to true self-driving cars.
For my framework about AI autonomous cars, see the link here: https://aitrends.com/ai-insider/framework-ai-self-driving-driverless-cars-big-picture/
Why this is a moonshot effort, see my explanation here: https://aitrends.com/ai-insider/self-driving-car-mother-ai-projects-moonshot/
For more about the levels as a type of Richter scale, see my discussion here: https://aitrends.com/ai-insider/richter-scale-levels-self-driving-cars/
For the argument about bifurcating the levels, see my explanation here: https://aitrends.com/ai-insider/reframing-ai-levels-for-self-driving-cars-bifurcation-of-autonomy/
Understanding The Levels Of Self-Driving Cars
As a clarification, true self-driving cars are ones that the AI drives the car entirely on its own and there isn’t any human assistance during the driving task.
These driverless vehicles are considered Level 4 and Level 5, while a car that requires a human driver to co-share the driving effort is usually considered at Level 2 or Level 3. The cars that co-share the driving task are described as being semi-autonomous, and typically contain a variety of automated add-on’s that are referred to as ADAS (Advanced Driver-Assistance Systems).
There is not yet a true self-driving car at Level 5, which we don’t yet even know if this will be possible to achieve, and nor how long it will take to get there.
Meanwhile, the Level 4 efforts are gradually trying to get some traction by undergoing very narrow and selective public roadway trials, though there is controversy over whether this testing should be allowed per se (we are all life-or-death guinea pigs in an experiment taking place on our highways and byways, some contend). Even worse, so is the concern surrounding the circumstances involving Level 2 or Level 3 vehicles that are being improperly touted as ostensibly being at Level 4.
Since semi-autonomous cars require a human driver, the adoption of those types of cars won’t be markedly different from driving conventional vehicles, so there’s not much new per se to cover about them on this topic (though, as you’ll see in a moment, the points next made are generally applicable).
For semi-autonomous cars, it is important that the public needs to be forewarned about a disturbing aspect that’s been arising lately, namely that despite those human drivers that keep posting videos of themselves falling asleep at the wheel of a Level 2 or Level 3 car, we all need to avoid being misled into believing that the driver can take away their attention from the driving task while driving a semi-autonomous car.
You are the responsible party for the driving actions of the vehicle, regardless of how much automation might be tossed into a Level 2 or Level 3.
For why remote piloting or operating of self-driving cars is generally eschewed, see my explanation here: https://aitrends.com/ai-insider/remote-piloting-is-a-self-driving-car-crutch/
To be wary of fake news about self-driving cars, see my tips here: https://aitrends.com/ai-insider/ai-fake-news-about-self-driving-cars/
The ethical implications of AI driving systems are significant, see my indication here: https://aitrends.com/selfdrivingcars/ethically-ambiguous-self-driving-cars/
Be aware of the pitfalls of normalization of deviance when it comes to self-driving cars, here’s my call to arms: https://aitrends.com/ai-insider/normalization-of-deviance-endangers-ai-self-driving-cars/
Self-Driving Cars And Pointing Fingers At Level 5
For Level 4 and Level 5 true self-driving vehicles, there won’t be a human driver involved in the driving task. All occupants will be passengers; the AI is doing the driving.
One aspect to immediately discuss entails the fact that the AI involved in today’s AI driving systems is not sentient. In other words, the AI is altogether a collective of computer-based programming and algorithms, and most assuredly not able to reason in the same manner that humans can.
Why is this added emphasis about the AI not being sentient?
Because I want to underscore that when discussing the role of the AI driving system, I am not ascribing human qualities to the AI. Please be aware that there is an ongoing and dangerous tendency these days to anthropomorphize AI. In essence, people are assigning human-like sentience to today’s AI, despite the undeniable and inarguable fact that no such AI exists as yet.
There is something else important that you need to know and it entails the significant and hugely crucial difference between Level 4 and Level 5.
Level 4 relies upon an important concept known as an ODD (Operational Design Domain).
Here’s the formal definition of ODD from the SAE J3106 standard: “Operating conditions under which a given driving automation system or feature thereof is specifically designed to function, including, but not limited to, environmental, geographical, and time-of-day restrictions, and/or the requisite presence or absence of certain traffic or roadway characteristics.”
In short, the notion is that a Level 4 self-driving car is devised to operate in a particular operational domain and the vehicle can only be driving within the constraints. I guess you could say it is a kind of carve-out. Rather than blankly stating that a self-driving car is able to pretty much go wherever, instead the emphasis is to clarify where and under what circumstances it has been devised to operate.
These ODD’s are entirely up to an automaker or self-driving tech firm to individually decide what it should be for their wares, such as saying that their self-driving car won’t work at nighttime and only works during the day, or works only in one city but not in another. Meanwhile, some other automaker offers a self-driving car that will work both during nighttime and daytime, but maybe it won’t work in the rain.
And so on.
This can be quite confusing to the public and the media too, namely that these ODD’s aren’t defined in a standardized way (see my column coverage about efforts underway to standardize ODD’s).
This means that if you see a self-driving car driving past you, there’s no immediate way to know what its scope consists of. Maybe it has been devised to go only in a 10-block radius of where you happen to see it, or maybe it can go across your entire state to other states and other cities. Perhaps it can work in the rain, but not in snow. You have no direct way of discerning what the ODD is.
It could be somewhat said that a Level 5 self-driving car has no particular ODD and encompasses all other ODDs, suggesting that a Level 5 can go just about anywhere, though that is a bit of an overstatement and technically imprecise. Here’s a useful quote from the SAE standard that helps indicate the Level 4 versus Level 5 capabilities of the Automated Driving System (ADS):
- Level 4: “Permits engagement only within its ODD”
- Level 5: “Permits engagement of the ADS under all driver-manageable on-road conditions”
Now that we’ve laid the stage appropriately, time to dive into the myriad of aspects that come to play here. Why are some calling for the abolishment of Level 5?
I’m glad you asked. Some assert that Level 5 is simply not attainable. Level 5 is a mirage. A false idol. Fakery. An aspiration that cannot be realized.
According to these contentions, automakers and self-driving tech firms are pursuing something that gives them and everyone else false hopes. The problem is compounded by using resources and attention that ought to be going toward Level 4, they would contend. But the race to attain Level 5 distorts what is being done. Instead of putting all your eggs into the ostensibly practical and level-headed Level 4 attainment, there is a leakage of valuable efforts toward garnering a Level 5 that isn’t going to happen.
Stop messing around and get your act together by going all-in on Level 4.
Furthermore, there are additional and quite serious concerns about providing a ready-made open door for deviousness by those that wish to distort reality, shamelessly exploiting the Level 5 formalized status. Shady characters can claim that Level 5 is their goal and showcase snippets of evidence that seem to suggest they are on their way to Level 5. They presumably could not do this if there wasn’t a Level 5 definition, to begin with. The very presence of the false idol opens Pandora’s box for those that wish to assert that whatever they are doing is for the goodness of ultimately achieving the acclaimed Level 5.
Regrettably, they would stridently argue, the Level 5 is firmly implanted in the standard. The solution then is patently obvious, they would say, namely, kick out that troublesome and overinflated Level 5 from the standard.
We would then all realize that Level 4 is where we ought to be looking. No more wasted time or money on that darned Level 5. Hopefully, everyone would gradually forget that Level 5 ever even was stipulated. Indeed, future developers might be shocked to discover that there had been a Level 5 and chuckle when the old-timers tell them that we used to think that Level 5 was possible. A hearty laugh and lots of headshaking ensue.
Now, before I provide the other side of the coin on this particular line of thinking, I’d like to mention that there are lots of variants of that aforementioned argument.
For example, one common variant is that Level 5 is not necessarily unattainable, and instead the qualm is instead that it isn’t attainable anytime soon. Thus, rather than making a blanket statement that we won’t ever achieve Level 5, this variant posits that it might be many decades or more away, and therefore for the foreseeable future it is entirely out of reach.
You might wonder why those in that variant niche would also go along with ditching Level 5. They seem to agree that Level 5 is potentially possible. Why not keep it intact?
Their logic is somewhat akin to the ones that say Level 5 is never attainable, namely that in the short run it is inadvertently acting as a distractor and can be deceptively used by those that wish to take advantage of its existence in the standard. Take out Level 5 right now, and maybe someday in the future consider bringing it back into the realm of possibilities.
Let’s now take a mindful glance at the primary counterarguments to this line of logic.
First, some would contend that you cannot un-ring the bell. Level 5 has already been specified and promulgated. Even if you got it out of the SAE standard, Level 5 would still be floating around and likely cited.
Some people would not realize that Level 5 had been removed. Too much history about Level 5 is lying around and easily found on the Internet. All that would happen is mass confusion. Is there a Level 5? Was there a Level 5? Why did it get removed? Is it coming back?
Sure, you could perhaps persuade enough insiders to get it removed, but society is not going to likely give up on Level 5. In addition, those devious people that you are trying to avert, well, certainly they would make sure that Level 5 continues to be alive. Perhaps even inventing some nifty conspiracy theories about why Level 5 was removed. That could stoke even greater interest in Level 5, one supposes.
Second, there is a position taken that we do not know for sure that Level 5 is in fact unattainable.
Speculation on the achievability is strong on both sides of the question. For those providing compelling reasons that Level 5 is someday attainable, they would believe it nonsense or worse to remove it from the standard on the basis that it is presumed unattainable. They would likely fight tooth and nail on that supposition.
Third, not having a Level 5 in the standard would be like omitting a vital part of the scale and render the standard horrendously incomplete, diminished, and inadequate, ergo undercutting any due respect or conformity.
This also dovetails into the aspect that Level 5 provides a lofty goal and inspires researchers, inventors, and everyone involved toward achieving something greater than might otherwise be envisioned. Some liken this to earlier postulating that we can get humans to the moon, which at the time seemed potentially improbable. Nowadays the equivalent might be getting humans to Mars. As a side note, the Level 5 opponents would say that’s an improper analogous situation and that you would need to say that we are aiming to get humans to some really far away planet well-beyond our universe somewhere in the multiverse.
Anyway, I trust you get the sense that there are those on the other side of tossing out Level 5 that would stoutly argue that Level 5 ought to remain in. The horse is already out of the barn on that, they would contend. Meanwhile, I believe that even the most ardent believers in keeping Level 5 would all pretty much be open to making refinements and adjusting Level 5, as indeed the standard is regularly being updated and revised.
There is another tack that some take in the rumblings about booting out Level 5.
This angle is a bit easier to unpack and alas is based on misleading or outright misunderstandings about the standard and Level 5. I say that this is unfortunate since it happens to simultaneously postulate a position that seems relatively untenable and meanwhile incorrectly portrays the elements of Level 5 as clearly stated in the standard.
Here we go.
Some have stated that their major issue with Level 5 is that it supposedly aims at having self-driving cars driving anywhere at any time in any manner whatsoever. This is portrayed as an eyebrow-raising slippery slope that leads inexorably to assured doom.
This is perhaps brought about by a mistaken comprehension of ODDs. The argument goes that since Level 4 has ODDs, it won’t get into dangerous driving environments, while Level 5, with no ODDs, could get itself into some knotty and dire binds.
Ugh, time to set the record straight on that.
The usual way this is portrayed is by asserting that a self-driving car of a Level 5 is going to proceed onto precarious cliffs while off-roading or in other highly dangerous settings such as driving around in the midst of monsoon, all of which is seemingly and undeniably bad.
Apparently, in contrast, a Level 4 would have a defined ODD that would prohibit this kind of reckless and wayward type of driving.
Sorry to say, that doesn’t compute.
Let’s briefly return to the definition of Level 5 from the SAE standard: “Permits engagement of the ADS under all driver-manageable on-road conditions.”
Take a close look at that wording.
Notice that the wording refers explicitly to on-road conditions.
The Level 5 definition specifically and unequivocally refers to on-road driving only. Period, full stop, drop the mic.
Allow me to quote further from the standard: “On-road refers to publicly accessible roadways (including parking areas and private campuses that permit public access) that collectively serve all road users, including cyclists, pedestrians, and users of vehicles with and without driving automation features.”
As I’ve repeatedly exhorted, the existing standard considers off-road driving to be out of scope.
Anyone that attributes off-roading to Level 5 is misstating the standard and improperly expanding what Level 5 is. That being said, you are welcome to argue about whether off-roading should or should not be included in the standard, and you are equally welcome to aid in fostering off-roading autonomous levels or something like that.
But please don’t misstate the standard as it sits today.
Next, take a look again at this wording from the standard. I’d like to point out that the wording refers to driver-manageable conditions.
That’s quite important phrasing. It intends to indicate that the driving action of the AI driving system would be adherent to what human drivers can manageably do when driving. The standard refers to the Level 5 unconditional or not ODD-specific driving as nonetheless still being bounded by that which is only also drivable by a human driver.
Here’s a further indication in the SAE definition: “Unconditional/not ODD-specific means that the ADS can operate the vehicle on-road anywhere within its region of the world and under all road conditions in which a conventional vehicle can be reasonably operated by a typically skilled human driver. This means, for example, that there are no design-based weather, time-of-day, or geographical restrictions on where and when the ADS can operate the vehicle. However, there may be conditions not manageable by a driver in which the ADS would also be unable to complete a given trip (e.g., white-out snowstorm, flooded roads, glare ice, etc.) until or unless the adverse conditions clear.”
Okay, the point is that the AI driving system is presumably going to be driving when the circumstances are warranted such that if a human driver was at the wheel would be able to drive in that setting. I’ll readily grant that this is still somewhat ambiguous. Some human drivers do stupid things and try driving in settings that are unwise. No doubt about that.
But if you are going to twist the standard and argue that it forces the AI driving system of Level 5 into driving in unwarranted settings, that’s a distortion and not the indication.
You could somewhat reasonably argue that there is a kind of macro-ODD that exists for Level 5, namely that a Level 5 self-driving car is only supposed to drive in driver-manageable on-road conditions. That might not seem like an ODD as is defined for Level 4, and it might seem like a quite open-ended ODD of a sort, but it nonetheless smacks of the underlying conceptual underpinnings that there are indeed conditions or restrictions associated with Level 5.
For more details about ODDs, see my indication at this link here: https://www.aitrends.com/ai-insider/amalgamating-of-operational-design-domains-odds-for-ai-self-driving-cars/
On the topic of off-road self-driving cars, here’s my details elicitation: https://www.aitrends.com/ai-insider/off-roading-as-a-challenging-use-case-for-ai-autonomous-cars/
I’ve urged that there must be a Chief Safety Officer at self-driving car makers, here’s the scoop: https://www.aitrends.com/ai-insider/chief-safety-officers-needed-in-ai-the-case-of-ai-self-driving-cars/
Expect that lawsuits are going to gradually become a significant part of the self-driving car industry, see my explanatory details here: https://aitrends.com/selfdrivingcars/self-driving-car-lawsuits-bonanza-ahead/
Many more twists and turns are available on this topic, but I’ll save those for another day.
Well, a quick taste of one seems especially noteworthy.
If the ODDs of Level 4 are supposedly going to keep Level 4 out of the potential driving troubles that are purportedly in Level 5, what is it that somehow magically makes the self-defined ODDs of Level 4 so attributable to safer driving?
The answer is that there is no such guarantee.
Here’s again the SAE standard definition for ODD: “Operating conditions under which a given driving automation system or feature thereof is specifically designed to function, including, but not limited to, environmental, geographical, and time-of-day restrictions, and/or the requisite presence or absence of certain traffic or roadway characteristics.”
One could presumably craft an ODD for Level 4 that is as dangerous or perhaps more so dangerous than Level 5, though we would not expect anyone to do this. The assumption by some is that ODDs are going to be nicely and neatly devised into rather tight quarters that are assuredly safe boundaries, which is what most of today’s efforts are seeking to do. That doesn’t though stop someone from doing otherwise. One would hope that this isn’t a direction taken, but it is conceivable.
Now that we’ve covered the ongoing jawing about whether to ditch Level 5, you get to be the judge.
Ought Level 5 be dropped like a hot potato from the standard? Should Level 4 reign supreme?
Some suggest that the insiders stoking such flames are just trolling everyone and are having a bit of fun. Though, there is an inherent downside to crying wolf, even in jest and especially when doing so on the sly. Others are said to be trying to provoke deep thoughts and get some renewed energy into postulating what the levels ought to be. If so, one supposes that at least it might be helpful to ensure that the facts are on par and suitably sketched.
What will happen next, you might be mulling over.
As Sun Tzu famously stated, it is prudent to ponder and deliberate before you make a move. Then again, he also proffered that to be quick is to survive and that if not quick then all shall be lost.
Copyright 2021 Dr. Lance Eliot