The easy availability of roads leads to more people driving since it's the more effective option, which in turn leads to more traffic. Eventually the system reaches equilibrium where it's busy enough that other alternatives are more desirable.
Time and time again predictions of apocalyptic traffic doom have failed to manifest when roads are removed. The traffic just goes away largely because the traffic was an artifact of the roads being there in the first place.
People are highly adaptable. They take the path of least resistance. If that path is a road, they take it. If you make the road bigger, it reduces resistance and becomes busier.
The only problem is cost. For a city to out build unmet demand they would need to get to add several dozen layers of road, one on top of another (that is bridges over bridges for 12 layers). Nobody would seriously propose building such a thing. If we had it the mobility of the city would allow people to accomplish more things in the city than they can today.
Would traffic flow become infinitely fast if we removed them all?
Conversely, would it become infinitely slow if we paved every square inch of land in sight?
It's almost as if such simplistic models are worse than useless, huh.
I'd say that it's more like taking models to extremes is useless.
If your doctor says that you're consuming too much salt it is increasing your risk of a range of diseases, you don't turn around and say "But salt is necessary for the function of life, I can't just cut it out entirely!". The answer, as with most things - is somewhere in between.
With our current set of transport options - a mix of transport options are going to be optimal. The specific mix is going to depend entirely on the area and traffic patterns.
Market demand, induced and otherwise, is a good thing, if the externalities are also addressed. That's basically Musk's whole business model.