It's precisely because they realise this that they focus down the premade's alliance 2v1, even going so far as to sacrifice their own game to stop the premade's win.
Initially, they play the game as they usually would: aim to win, combining the winning of objectives with gaining kill points through PvP. Admittedly, in many random teams the objectives tends to be considered the stronger priority and a number of excellent attack opportunities would be missed because of the objective-focus or general avoidance of 'unnecessary' (they think) PvP.
However, as the game progresses, it becomes apparent that at any moment in which the premade is left alone to 1v1 either team, a wipe will be guaranteed to the random team and a big point and BH gain to the premade's alliance. Even in the event that one of the random teams was particularly combative and sought the opportunity to try to pincer the premade during one of these attacks, the effect will be somewhat limited - the premade's whole alliance will already have a substantial BH advantage from their first two attacks on unawares targets, the random team will not have that benefit. Some of the premade's alliance will likely become victims and the premade's DRK may be caught depending on the availability/use of their LB, but the premade DRGs and RPR will be long gone thanks to BH buffs, DRG speed buffs, and RPR teleport. Put simply: their main damage source remains buffed and safe and you're totally reliant on the other random team doing the same when the premade regroup and come to attack you, which in random matches, is incredibly unlikely to happen.
This isn't to mention the obvious difference in coordination between random teams and premade ones - even with a well coordinated random team, damage will be slightly spread as players cast at slightly different times, giving chance to Guard/Heal/Run. In any decent premade, this damage is almost instant.
The chances of a random team having a random DRK/RPR/DRG all reach the same coordination as a premade on voice chat?
So yes, you can try to counter them and some teams will, but the result will not compare to that of their attack on you.
Eventually, even to the most beginner players, it becomes obvious that the points being endlessly farmed by the premade are creating such a drastic score difference that to leave it longer will only result in the premade win. Some will resort to defeatism ('just end it fast'), some will try to push for second place and will try to leech what kills they can from the premade's attacks on the other random team, others will "look for an opportunity to work with the other team" and form a silent agreement with them, resulting in both teams practically camping near to the premades respawn while the objectives go to whichever random team is in second.
(And before you say 'camping outside of base isn't the way', as I have mentioned above, doing a pincer attack will have limited effect too - the premade will wipe the opposite team in seconds before running away to regroup. Your own team will have to be in exactly the right place at exactly the right time. Their attacks will have to be just as simultaneous, with the same meta jobs used by the premade. In a random team where communication is fully in-game with a team of varying skills and experience, this will not happen. Hence many resort to having both teams simply camping outside of base. Not ideal, but at the end of a desperately imbalanced game, at least they know it works when nothing else for them has)
Casual players' avoidance of PvP (unless it's directly over an objective) doesn't help the situation, I agree. But it can't be denied that premade teams have a significant advantage over random teams, not just through game strategy, but through the availability of a very closely coordinated group of meta jobs. Hence why they play as part of this pre-formed party of friends, rather than queue solo. The main issue is that this advantage is so great that it removes all competition from a game and helps to push disinterest in PvP amongst new/casual players.