Sure it's artificial, so is everything rules wise in ASL. When did you last look out your upper story window and see the faint outline of hexagons tracing through the road and into your yard? Maybe once or twice after a 6 hour game, but I bet they were not there the next morning.
While ASL is a tiny bit more sophisticated than knocking over Airfix soldiers with your index finger going "rat-tat-tat", all games have rules to give some faint illusion of reality. Most war boardgames use hexagons (a few use squares and more, still a minority, use irregular 'areas') because that type of regular polygon tiling introduces the least distortion when counting relative distances.
Once you use hexagons then you have two choices, either use the painted obstacle outline for LOS obstruction or assume any obstacle occupies the whole of the hex for rules purposes.
If the later is the only type (like ASL inherent terrain) then LOS can be determined by simple geometry without any explicit reference to the hex centre (though the effect is actually equivalent to from hex centre to hex centre).
For the first type, which is the dominant ASL type (though there is some complete hex types) then you have to lay down a ruler, thread, dick or laser to see if the artwork blocks LOS. For SL (in its design stage) it was decided to use the hex centre dot for LOS rules purposes. That centre dot is the 'centre of mass', as it were. Geometrically if you draw a line through a hex (or square) from any point outside the hex the longest possible line segment that is within the hex boundary will pass through the centre. In addition any such line that goes through a hex centre exactly divides the area of a hex into 2 identical (in shape, size and area) parts.
The game assumption is that going through the centre will contain the biggest portion of the hex. That may be or may not always be true (you may be able to see 80% of a hex but a hut blocks the centre dot), but it allows the rules writer to lay down a solid rule and avoid the "but it just touches the hex corner" argument when 99.9% of the hex is masked. You still will get the occasional centre-centre LOS just about brushing an obstacle, but at least in that case you usually will be able to see 50% of the hex.
While the C-C method is not perfect it does allow for a compact, relatively shark proof, rule method of maximising the hex area that is visible. As RobZagnut has pointed out there are other methods that work (quite?) well. Using a concept (centre dot) that all except complete cret... sorry ... severely mentally challenged have an almost instinctive grasp of being the best chance of 'hitting the bulk' of the hex helps in the absorption and acceptance of the rule.
Will the C-C rule always ensure that if a legal LOS exists that you can see, say, >50% of a hex? No. It does maximise the chances of that. It is a simple, easy to understand method, can be checked with reasonable accuracy and is fairly sleaze proof. And it is in The Rulebook, Amen.