Mental wayfinding, key for humans, likewise for AI.
Do you know where you are right now
The odds are that you do.
I don’t mean that you perhaps know that you are at a particular street address or sitting in your favorite chair at home, but instead, I am referring to the aspect that you spatially know where you are.
Via both your senses and your mind, you might realize that you are ten feet away from a nearby TV and that you are five feet away from the nearest window. As you sit inside a room, your mind has an abstract model that keeps track of where you are and where other nearby objects reside.
On top of that realization, your mind also knows that the room is within a building, the building is within a neighborhood of buildings, and the neighborhood is within a city, which is within a county, and within a state, and within a country, etc.
Imagine using something like a mental version of Google Earth and having your mind be able to zoom in and zoom outward, quickly imagining your position from a faraway location and at the same time being able to get close-in and know exactly where your feet are placed and your immediate and within reach surroundings.
This knack of cognition is typically referred to as mental wayfinding.
Humans seem to have a powerful capability of mental wayfinding.
To set the record straight, wayfinding is not solely a function found within humans and decidedly is an inherent facility of many living organisms.
We do though seem to take this wayfinding to a higher-level of thinking.
Here’s what that means.
The notion of spatial orientation and spatial elements permeates a lot of what we do, far beyond the mere act of locomotion.
Sure, you use your mental wayfinding when trying to find the kitchen or while hiking in the woods and aiming to get back to your campground, but this is only a sliver of the wayfinding usage.
Wayfinding and spatial processing are subtly found in how we describe a lot of what we do.
You might describe a longtime friend as being a near-and-dear close friend and bring up reference to your distant aunt. Notice that the words “close” and “distant” are conventionally about spatial matters and would normally refer to physically being near to someone or something. In this alternative use, those words are more so about the relationships that you have with those people.
In short, we borrow lots of spatial meanings to be used in various other significant ways.
Why so?
One theory is that our minds are so immersed in spatial capabilities that it wants to try and use that built-in aptitude in as many ways as seems possible. Presumably, we might as well leverage a handy feature that already is ingrained in our minds and use it to our survival advantage. Spatial processing is akin to duct tape, namely, it has a lot of uses and we keep finding new ways to take advantage of it. We can’t help ourselves.
Let’s then separate the use of spatial processing from that which aids our ordinary physical navigation and consider how spatial facets are used in more abstract and unheralded purposes.
As an aside, the traditional assumption is that spatial processing is intended principally for locomotion and movement within our physical world, a seemingly obvious point, though some try to argue that maybe spatial processes were intended really for our higher-level thinking and just so happened to get harangued into being used for bodily navigation. A fascinating proposition, though one that is hard to make as strong a case as being more likely the other way round.
Well, at least we do know that wayfinding is existent and quite exquisitely useful.
Consider your mind to be filled with knowledge and those gold nuggets are arranged somewhat like an everyday library or bookstore. When you go into a physical library or bookstore, you might know that the books on cooking are on aisle seventeen and that the books about car repair are stacked toward the back of the store.
Likewise, we often find ourselves trying to think about a subject or topic and roll our eyes, as though we are peering inward into our minds, seeking to find that location of an obscure fact about a great historical explorer or maybe trying to remember an algebraic equation we learned in grade school.
It seems as though we use wayfinding in how we store information within our minds.
The same spatial memory that remembers where you left your toothbrush is also able to remember the notable fact that Einstein developed the theory of relativity.
A toothbrush is a physical object in the real-world and has some position thereof, while the snippet of memory about Einstein is a piece of knowledge that conveniently is organized and stored inside your head, yet they both have in common the use of your cognitive spatial powers.
One can persuasively argue that mental wayfinding is a crucial cornerstone for humans being able to think.
Perhaps the aspect that we seem to be able to extend thinking far beyond that of other animals is that we have taken spatial processing to a heightened place. This does not suggest that wayfinding is the only basis for the incredible reaches of human thought, and merely emphasizes that it is seemingly a vital reason for the amazing aspect of human intelligence.
Why all the bother about this?
For those embarking upon crafting AI, the underlying vitalness of mental wayfinding needs to be given its due respect and attention.
One might say that wayfinding has earned such a right.
Not everyone necessarily sees things that way.
Time to back up and provide some context.
AI is intending to achieve artificial intelligence that presumably is the equivalent of human intelligence.
The vaunted quest involves grasping how human intelligence works, doing so to try and give us insights into what artificial intelligence is most likely to also need.
Some within AI are apt to argue that we do not necessarily need to deconstruct or reverse engineer human intelligence and instead can construct whatever we want, as long as outwardly it seems to showcase the same semblance of intelligent behavior.
That indeed is one means to approach the problem and lessens the difficulty of needing to crack open the inner secrets of how the human mind works. Don’t worry about the human mind and just proceed ahead to make something that appears to have the same result.
Others believe that the treatment of the human mind as a kind of an impenetrable black box is a false path and ultimately doomed to failure. Rather than trying to devise the equivalent of a mind from the ground-up, the hope is that by figuring out the wonders of the brain we will be more likely to arrive at AI.
If you believe that the open-the-brain is the appropriate pathway to AI, we seem to already know that mental wayfinding is instrumental to human intelligence, and therefore logically we ought to be devising and implanting such wayfinding into AI systems.
For those on the other path, wayfinding offers some interesting considerations but does not loom as large as an essential component to be toyed with.
Some applications of AI readily make wayfinding a sensible and needed ingredient.
Consider this intriguing question: Should AI-based true self-driving cars embody a type of mental wayfinding, not just for physical navigation, but also as a general knowledge-based model for operation and driving?
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 a 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 a 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 point out, see my indication 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 Mental Wayfinding
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.
First, let’s cover the apparent aspects of physically-oriented navigational wayfinding.
Using GPS and other navigational aids, including the IMU (Inertial Measurement Unit, see my explanation at this link here), the AI driving system has to keep tabs on where the vehicle is and where it is going.
Also, similar to how robots navigate, the AI driving system makes use of SLAM (Simultaneous Localization and Mapping, which I describe at this link here), a computational technique that enables the AI to do physical wayfinding.
Once AI self-driving cars become prevalent, humans will presumably no longer need to be especially concerned about wayfinding when traveling via an automobile since the AI will take care of that chore for them.
No more needing to consult a map before a journey or during your travels. Just sit back in the seat, maybe reclining as it is anticipated that self-driving cars will have seats that convert into beds, allowing for taking a nap during a lengthy ride, and enjoy the trip.
In the United States alone, we drive about 70 billion hours annually and need to keep our minds sharpened toward the driving task and the navigational considerations too. Of course, in modern times you can merely follow the GPS step-by-step instructions and not need to mentally have a larger image in your mind of where you are going, yet nonetheless, you are still exercising some amount of mental wayfinding.
Some worry that once we become reliant upon self-driving cars, our mental wayfinding related to navigation will wane.
It is potentially a skill that if not kept up-to-date will atrophy.
You might be thinking that it won’t matter if we do weaken our physical wayfinding prowess since we will have those self-driving cars to do our bidding.
The rub consists of the potential leakage due to such a weakening.
One theory holds that if we begin to lose our physical wayfinding mentalism, we will see a similar degradation in the other realms of our mental wayfinding.
Trying to remember the name of that famous historical explorer or the fact about Einstein is going to become harder and harder for us humans to do. The belief is that the wayfinding as an overarching capability will diminish and therefore spill over into all other mental uses of wayfinding.
What do you think?
Will AI self-driving cars inadvertently lead to the human mind becoming less sharp and maybe we will become dumb and dumber?
It is a twofer, namely, we lose the ability to do wayfinding in the physical world, plus we undercut our mental prowess that relies upon wayfinding.
Here’s a means to thicken the plot and make things even worse.
Some believe that we will opt to no longer walk as much, due to the advent of self-driving cars. Instead of walking down the block to visit a neighbor, we will hop into the handy-dandy self-driving car and have the AI drive us there (for more on this social phenomenon, see my discussion at the link here).
If you go along with that theory, humans are possibly going to exercise less and potentially plump up.
The totality of this grand convergence results in humans becoming fattened and slovenly, along with losing their mental edge and becoming mental zombies, all as a result of the convenience and marvelous addition of self-driving cars into our world.
Wow, a lot of baggage seems to be piled on top of those self-driving cars.
Anyway, let’s shift gears.
Can AI use mental wayfinding in ways other than purely for navigation?
Yes, absolutely.
One such approach involves the use of knowledge graphs, considered an AI-related technique that structures knowledge into a representation that might be construed as a tree that branches and stretches out in a multitude of directions.
Ponder how this might be used in an AI-based self-driving car.
Suppose the AI is driving the self-driving car and getting near to railroad tracks.
You might not normally give much attention to railroad tracks while driving a car. The surprising aspect of railroad tracks is that nearly 500 deaths per year occur in the United States due to failing to safely cope with a railroad crossing. There are about 128,000 public railroad crossings and approximately 180,000 miles of railroad track in the U.S. (for more about the dangers for self-driving cars and railroad track crossings, see my analysis at this link here).
The AI could treat each railroad crossing as an individualistic and prior unseen type of driving circumstance. Or, the AI might have a treasure trove of aspects previously recorded when driving over railroad tracks, stored as part of its knowledge-base about driving.
Via the use of a mental wayfinding component in the AI, perhaps based on a knowledge-graph, access is made to those various techniques and possibly even prior experiences of having crossed railroad tracks.
Those are then fused and utilized when approaching a railroad crossing that the AI has not previously encountered.
The AI-based mental wayfinding can be used in a wide variety of other contexts, including carrying on discussions with the passenger inside a self-driving car.
Whereas today’s self-driving cars have a quite limited Natural Language Processing (NLP) capacity, in the future it is anticipated that the NLP will be able to engage in rather wide-ranging dialogues with riders and possibly even offer a kind of therapeutic counselor as both your driver and emotional adviser (for my explanation about this, see the link here).
Conclusion
Do you ever let your mind wander?
Are there times at which your thoughts seem to amble, take a stroll, go for a walk, or otherwise embody a spatial metaphorical kind of travail?
AI might need that same spatial set of undertones and capacities, without which we might not end-up with the aspirational AI and instead be laden with a lesser and non-spatially oriented type of AI.
I’m voting that we keep AI on the invigorating path toward ingratiating the marvels of mental wayfinding and spatial wonderments.
As Einstein was known for saying, look deep into nature, and then you will understand everything better.
