I don't think so. Pros are already pretty good at the endgame. Yeah, they make blunders now and then, so it's possible that AlphaGo would gain a few points in the endgame in some games, but not enough to overcome a significant lead and not every time.
In any case, I personally find it more interesting to see what we can learn from AlphaGo about the opening and the midgame.
My writing was sloppy, let me rephrase it: in a pro match, whatever is the estimated score when the game goes into the endgame, that will likely be the final score.
Where I wrote that a pro will blunder "now and then" I should have written "rarely" -- I don't have data to back this up but I'd guess once in every n games for some n > 30.
In any case, I personally find it more interesting to see what we can learn from AlphaGo about the opening and the midgame.