I don't see how you can term them as "clear mistakes" when the game is playing at a higher level than any of us meatbags. In this case (according to DeepMind) it's no different to a racing driver backing off the pace in the last few laps if they have a big lead - it's better to guarantee a win than to win by a large margin.
You might be confusing correctness with relevance. You can pass at the last move and lose 1 point, but if your lead is 5, its the same win-result but not the same count-result. Well, that would be a mistake by all accounts, even if AlphaGo made it.
Some of the moves AlphaGo played both in this game and the previous one are very definite mistakes, but they were irrelevant to the difference in the game.
> Well, that would be a mistake by all accounts, even if AlphaGo made it.
It is only a mistake to your human biases. AlphaGo literally has no conception of the margin of a win. It doesn't care either way how many points it wins by, as long as the win percentage is maximized.
Exactly. It's actually a pretty common behaviour for tree-searching game playing algorithms. Unless they're set up to explicitly give weight to faster wins or higher margin wins, they get into an unassailable position and then just kind of dick around. If every move's a win then it doesn't matter which move you make, right?
I think that his point is that it would be interesting to see AlphaGo pushed to its limit and playing a point-maximizing game. Which would be pretty terrifying.
