San Francisco with a Victorian home cruising down the street on just another ordinary sunny day.
Hearst Newspapers via Getty Images
Sometimes, you can be quite surprised by what you might see on the sunny streets of San Francisco.
For example, this proverbial City by the Bay is a Silicon Valley favorite right now, allowing an abundance of self-driving cars to crisscross the steep hills and teeming streets, doing so in a visionary flurry of experimental tryouts for emergent driverless cars (for my coverage see this link here).
If that isn’t enough of an unusual scenery for you, there is something else that San Francisco also is known for, namely houses that from time-to-time are rolling throughout the avenues and keep afar from the fabled cable cars. Though some have jokingly said that these slinking along homes have a mind of their home, the reality is that there is a long history of putting treasured houses up on giant dollies and moving them to other parts of town.
To be clear, there are humans at the wheel of those moving homes (not driven by ghosts and nor any venerated AI-based driving systems).
In this latest case, a picturesque Victorian home that is popularly known as the Englander House was moved to make way for a proposed apartment complex. One supposes the easiest path would have been to simply demolish the now outdated and somewhat dilapidated house, crushing it to the bone and carting away the remains. Instead, the more than five-thousand square-foot structure was carefully and gently rolled to a new spot, which though admittedly only six blocks away, nonetheless constituted a heck of a distance to cart an entire two-story home.
By lifting the home onto rollable wheels, the affair became one of trying to navigate the winding and hilly streets of famed San Francisco, known as being one of the hilliest cities in the world. Fortunately, no need, in this case, to try and shove the house across the Golden Gate Bridge. The local streets that would witness the inching along house were prepared beforehand to make way. This included cutting back overstretched tree limbs, removing parking meters, temporarily taking down road signs that were potential obstructions, and the like.
You might be wondering whether the posted speed limit signs were also removed, and in which case, could the house opt to sprint its way to the new location. Sorry to disappoint, the speed limit still applied, though it wasn’t a consideration since the rolling process had a top speed of about 1 mile per hour anyway. No sense in laying out rubber on the roadway from doing any kind of tire screeching starts or donuts, and just make sure that the house safely reached its ultimate destination.
Here’s a question for you.
Suppose you were driving your car in San Francisco on that particular day, and also assume that you did not know beforehand that a house was being moved. In that scenario, you would have been quietly driving along, minding your own business, and upon turning a corner on some ordinary street there would suddenly be looming ahead of you a house.
Mind you, not a house that was on the sides of the street and sitting still, nailed to the perch, but instead a home smack dab in the middle of the street. Perhaps you would end-up going head-to-head with this menacing figure. In a street duel involving going face to face with a house versus a car, consider the likely winner of this driving battle. The house weighs a lot more and has quite a hefty size advantage, while your car is nimble and can make circles around the snail-paced home.
I’d bet that most drivers would have let the house win.
The odds are that you would have been caught off-guard at seeing a house coming down the street toward you. Since this was happening in slow motion, you would have plenty of time to take in the sight and mentally calculate what was taking place. Other than a mild surprise and perhaps giving you a bit of a chuckle, it seems likely that you would have taken the matter in stride.
That being said, it is not every day that you manage to observe a house in the middle of the roadway and that is moving along too. Someone visiting San Francisco as a tourist might immediately assume it was a touristy act being undertaken by the authorities that run Frisco, possibly attempting to gain renewed attention to the Golden Gate City. I suspect that it would be easy to shrug off the experience and maybe make a posting on social media in a nonchalant manner, doing a trendy insta-bragging or a so-called humble-brag about what you just saw.
The odds are that you would have made a U-turn and made your way beyond the reach of the iceberg-slow moving house. Of course, the reality is that you probably would not have gotten especially close to the being-moved home anyway. There were plenty of barricades and road closures, along with bright and flashing warning signs and police enforcing the right-of-way for this Victorian-style masterpiece.
All told, a human driver would have presumably remained calm and at most been curious about how a house managed to get onto the street. Despite never having seen such a roadway condition before (how many of us have ever experienced a rolling home coming down the street?), a person witnessing such a spectacle can readily make a mental sense of what they are seeing.
It is a house. It is on the street. It is slowly moving. It poses no immediate danger. I can drive away without difficulty. That’s the extent of the driving mental contortions, and the remaining cognitive action can entail the why and how of the story behind a house confronting you while on a normal driving journey.
Shifting gears, the future of cars consists of self-driving cars. The AI-based true self-driving cars are driven entirely by an AI driving system. There isn’t a human at the wheel, and in fact, the expectation is that most self-driving cars will not provide any provision of human-accessible driving controls. The AI will be the sole driver and no human will directly be involved.
This gives rise to today’s intriguing question: What would a self-driving car do when suddenly facing a house that was moving down the middle of the street?
Let’s unpack the matter and see.
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 (see my explanation at this link here), 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, see my coverage at this link here).
Since semi-autonomous cars require a human driver, the adoption of those types of cars won’t be markedly different than 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.
Self-Driving Cars And Those Rolling Houses
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 today’s AI 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. For example, current AI lacks any semblance of common sense and nor has any kind of common-sense reasoning (for further background, see the link here).
This is important to keep in mind when considering the matter of a house coming down the middle of a street on a driving trek.
Humans comprehend that the object ahead of them is a house. Houses are usually not in the middle of the street. Houses rarely move, other than perhaps shifting here and there due to earth movement or seismic activity. Overall, a human would readily grasp that this is an unusual situation and would realize that this Victorian house is in the wrong place and doing something it should not be doing, namely cruising along on a presumably active street meant for cars, bikes, motorcycles, vans, and the like.
The house is altogether out of place. A human would reason that there must be some logical explanation for this oddball circumstance. How could a house get into the street? It is not as though a house is like a pet dog that got off its leash and sneaked out to play in the middle of the roadway. A house is not a dog. Houses do not customarily of their own volition opt to stray.
A quick visual scan of the driving scene would undoubtedly showcase the special roadway and moving crews that were shlepping along with the house. The street would be barren of all of the usual trappings that you’d expect to see on a typical everyday roadway. Thus, even if you somehow did not notice the blizzard of warning signs and somehow avoided getting stopped by those esteemed guardians securing the route, you nonetheless would reason about what you are seeing and be able to turn the extraordinary setting into something sensible and understandable.
What would an AI driving system do?
As mentioned, today’s AI is not sentient. It cannot “reason” in the manner that humans can. Let’s trace the technological aspects of what might occur.
The odds are that the sensors would detect that things were amiss. Self-driving cars are loaded with various sensors such as video cameras, radar, LIDAR, ultrasonic devices, and the like. You can certainly assume that all of those sensors are pretty much going to detect a house that is sitting in the middle of the street.
Realize though that the various sensors each have their own range of detection. This means that some of the sensors would sooner detect the object than others. Within the AI driving system, there is a function or component that undertakes Multi-Sensor Data Fusion (MSDF). This capability is programmed to compare and contrast the results of the sensor data collections and try to resolve any differences and also amplify any similarities. The visual imaging interpretations would be indicating that a large object is blocking the roadway. The radar and LIDAR would likewise be reporting that a large object is straight ahead. Etc.
The AI driving system maintains a virtual world model, essentially an internal frame of reference about what the outside world of the surrounding driving scene seems to consist of (for my analogy to the block’s world of Minecraft, see the link here). Based on the sensor data and the algorithmic interpretation of the sensor indications, the AI driving system likely would make a mark in the virtual world model that something rather monstrous in size is blocking the roadway ahead.
Notice that I have carefully worded that description to refer to the thing as an object and not refer to it as a house.
We don’t know that the AI driving system would necessarily classify the massive object as a house per se. For some AI driving systems, a house is considered an object that sits outside of the roadway. It could be the case that the AI driving system would be trying to compare the house to objects that are usually found within the path of the street, such as other cars, trucks, moving vans, pedestrians, and the like.
The AI doesn’t “know” that a house is a house, at least not in the manner that you and I understand the meaning of the word. Any programming related to the object that is being called a “house” would likely have simple attributes associated with it. There is no semblance of realization that a house can have a family that grew-up within the home and lived a wonderful life there, enjoying the company of others and the raising of children, and so on.
Consider for a moment all of the memories and thoughts you might embody about the nature of houses and homes. Essentially, none of that robust wealth of knowledge and understanding is within the reach of today’s AI systems.
Okay, we hopefully all agree that the AI driving system is unaware of what a house per se is (and, please avoid anthropomorphizing AI). At this juncture of the step-by-step exploration, the AI driving system has detected a large-sized object, taking up the majority of the ahead driving scene. This is beyond the size of a tumbleweed. It is beyond the size of a piece of furniture that might have dropped off the back of a truck.
All told, the odds are that the AI driving system would feed this indication about a colossal blockage up ahead into the portion of the driving system that plans what to do next. The routing algorithm would potentially have various digital maps and would refer to the maps. If the street ahead is unavailable for traversal, the AI driving system would be seeking to identify an alternative path.
We can assume that the self-driving car would seemingly aim to make a U-turn, and head some other way, utilizing other streets to avoid getting blocked by the object on this particular street.
In that case, consider this matter from the perspective of a car. A human driver would likely have made a U-turn and found some other path to proceed. An AI driving system would also have likely made a U-turn and opted to take a different route to proceed on its driving journey. If you were standing on the street and watching a car that came up to the situation, you would see the car make a U-turn and try to navigate away from the house coming down the roadway.
Could you immediately discern whether the car was being driven by a human versus being driven by an AI driving system?
Sure, if you could see into the car, you would obviously have seen a human sitting in the driver’s seat in the case of the human-driven car, and you would presumably have observed that there wasn’t anyone sitting in the driver’s seat of the self-driving car. In that manner, you could readily tell the difference.
But suppose you couldn’t see into the car or the car was far enough away that you couldn’t discern the nature of what might or might be inside the car. From the external behavior of the car, you would not seemingly be able to discern whether it was being human-driven or being driven by an AI driving system.
Is that good or bad?
Well, you have now reached a vociferous question that many are scratching their heads about.
On the one hand, yes, the AI driving system seemed to do the same thing that a human driver would have done. That seems reassuring.
Per my aforementioned depiction about the AI driving system, you undoubtedly realize that the AI didn’t know anything at all about a house or a home, at least in comparison to a human. In this specific setting, it would seem that not “knowing” about houses or homes did not make a notable or discernible difference. The resulting driving actions were the same.
Not all driving actions are of the same ilk.
There are driving actions that entail human thinking that AI driving systems are not yet able to attain, and for which it is unclear if such AI will ever be possible. As such, the AI driving systems are going to be quite rudimentary and lacking in the depth of comprehension that humans have and that humans seemingly leverage when driving a car.
Some worry that as a society we are going to piecemeal allow ourselves into having self-driving cars on our roadways that seem or appear to be driving as though a human was at the wheel, and yet the AI driving system is not even in the same ballpark as human cognition.
The other side of that coin is that humans have all kinds of human foibles that come to play when they are driving. There are about 40,000 car crash-related fatalities each year in the United States alone and approximately 2.5 million related injuries (see my listing of crucial driving stats at this link here). Much of those statistics are due to drunk driving and distracted driving. The AI driving systems won’t be driving while drunk and will not be watching cat videos when driving the vehicle.
Are we willing to trade off the advantages of AI driving systems over the foibles of human drivers, though also realizing that the AI is not sentient?
We will need to accept the notion that the AI will be of a merely mechanical and programmatic capability, and overcome a false sense that the AI is of caliber with human thinking and intrinsic human capabilities.
You might have been surprised to discover that the mere presence of a slowly rolling house in the middle of a street could spur this kind of illustrious discussion.
Turns out fortuitously that the quirkiness and mystery of a house existing where it is not expected serves as a marvelous foil upon which to reveal the oft unspoken aspects about AI and the capabilities of today’s AI driving systems. The day-to-day acts of self-driving cars starting to roam our streets can inadvertently lull us into assuming that AI driving systems are the same as human drivers, embodying the same cognitive capacities, and meanwhile excising the adverse role of human emotions and human faults that occur while driving a car.
There is no free lunch.
The AI driving systems are limited in what they can do. A tough tradeoff is going to occur as society has to decide between the advantages of self-driving cars versus the advantages of human drivers. If driving a car is relatively routinized, and some sufficient range of driving scenarios are programmed to be handled, this might be enough to weigh the scales toward those AI driving systems and avert the downsides of human-driven cars.
Time will tell.
Meanwhile, I am determined to put some sturdy wheels onto my house and take it for a spin, including refining my homegrown AI driving system to be able to drive a house rather than a car. One thing seems pretty sure, my AI driving house can stomp on any of those pesky AI self-driving cars, in a heartbeat.