The reality is there are a lot of impaired drivers on the road. May be they have been drinking, have early onset dementia, or just a mixture of prescription medications with undocumented side effects. A lot have spent half the time sending text messages on their phones. Whatever the circumstance, some of the tools we have now let them survive longer or at least pretend they know what they are doing.
Those stories make terrible clickbait unless there's a GPS involved, then the GPS can take 100% of the blame as if there was nothing else involved, and it makes itself insta-clickbait.
A cognitive map is an internal mental map of how some region of space is connected. Using landmarks and signs to follow a route from a map naturally builds mental connections between your observations of the world and the corresponding points on the map, in a way that following spoken directions does not.
Here in Melbourne we have (shortly: had) the Melways. A4 size traditional map-book, two pages covers a good couple of suburbs. Let's call it a 5km-wide area.
When you look at this map, you see the relationships of suburbs. Balaclava is just to the east of St Kilda, which flows to the south to Elwood, etc. You see the major arterial roads and where they lead. You build a proper, high-level mental model of where you are.
Contrast this to the GPS. "Turn left in 100m" tells you nothing of your locale. It gives you no context. You see a sign that says "Elwood" and you're heading south because the beach is to your right and you know you want to be in St Kilda but this isn't useful to you: your brain doesn't instinctively think, hmm, that just feels wrong.
GPS is amazing until it doesn't work, and then you are truly lost.
I tend to use the GPS on my phone the first couple of times I need to go to some particular place. After that, the experience of having driven the route a couple times is good enough that I can find my way.
But then, I'm one of those people who generally only needs to see a route once in order to know it. When I visit people in other cities, after a few days I'm often showing them how to get to places rather than vice versa.
An anecdote: I've been driving in this one particular city for ages, I know the streets and their properties - yet while I was busy with traffic (rain and night and pedestrians, oh my), it only took a "turn left NOW" command to send me wrong way into a one-way street (instead of the correct one, a block down). I knew that specific street was one way, but not focusing on navigation at that moment, I have blindly obeyed, as it looked somewhat reasonable; only after turning did I realize what I have done (fortunately, nothing else came of that incident). Note that I do not have blind trust in algorithms - quite the contrary (turned out the navigation's data was wrong), yet I have outsourced a part of my cognitive load onto the device.
I'm not blaming the device: the faulty action was all mine (that's what you get for a talking toaster ordering you around). What I got was some insight: I no longer wonder why someone with more trust in their computing devices could do something so "blindingly stupid." This is probably the one area in UX that's hard to fix: the assumption that Friend Computer Is Always Right.
It is very difficult for me to imagine driving 250 miles out of the way (especially when I'm tired and really want to get where I'm going); or driving into the ocean; or driving off a bridge because I've ignored all the road closed signs. I don't even shut my brain off to this degree when another human is helping navigate.
The article is saying our cognitive maps are deteriorating because navigation software provides the information instantly at hand. For some people that may be true. I've noticed this effect in other aspects of my life, though not with navigation. But even if true, that does not explain people ignoring all other indicators that something is not correct.