They have to be exaggerated or you would see no relief at all.

As a rule, changes in vertical elevation are tiny compared to the geographical extent of morphological features. For example, on earth, the greatest difference in relief is 14 miles, from the bottom of Challenger Deep in the Marianas Trench to the top of Mt Everest in the Himalayas. But the earth has a radius of about 4000 miles! A scale model of our planet say, the size of basketball, would feel as smooth as a ball bearing to us. The largest and most dramatic morphological feature of the planet, the abyssal plains, are about two miles deep, and very little of the continental land masses is anywhere near that high above sea level. Earth is pretty flat, its a big deal for us only because we're just a couple of meters high.

Photographs of the moon show these spectacular landscapes, but they are always taken when the sun angle is very low and shadows are long. In reality, the most rugged mountains on the moon are probably lower and smoother than the Appalachians. A few simple measurements would have shown this to be true, but until we actually traveled to the moon we always assumed the relief would be jagged and extreme, like new, unweathered mountain ranges on earth like the Alps or Andes. I remember astronomy textbooks telling us to expect jagged cliffs and steep slopes on Luna because there "was no weather to erode the hillsides".

We were also led to believe that Mars would be flat and featureless, highly weathered like Arrakis. Von Braun and Ley imagined our winged spacecraft landing on vast sand seas there using skis and skids. Instead, Mars looks more like west Texas.

In aerial photo and topo mapping the vertical scale is always exaggerated as much as possible, because the users of 3D imagery are interested more in the 3rd dimension for scientific or engineering purposes, and because the technology will support greater precision in elevation than it will in x and y. In drainage studies, for example, a change of elevation of several cm over a kilometer can be critical, so if you can see it, you go for it.

So its not just a question of aesthetics, but a question of how the cartography will be used and the technology that is used to develop it. The one factor where aesthetics and "art" does take a role is in the illumination of shaded relief maps. They are always drawn as if the sun is in the NW, which it never is except in late afternoon in the Southern Hemisphere. Any other illumination direction and the map shadows just don't "look" right; craters look like domes, and valleys look like ridges.