Suggestion for a new kind of training

I was reading an article in which Carlsen was asked: "how many moves, in advance, can you see?" and he answered: "I can see many moves in advance, but the most importan thing is the capacity to understand the strategy value at the end of the calculation".

Sometimes when I see a game, I'm not very skilled to understand the strategic value of a position. But computers are very good at this. My suggestion is a kind of puzzle (I don't know if it exists anywhere) in which a certain position is shown and players are asked to answer if the advantage is for white or for black. What do you think about it?

I like the idea, but the comp. does not alway tell the truth. It judge based by what the programmer set it to do. F.x. doubled pawns is a weakness ( insted of 1 point each pawn that is doubled, it may say 0.8 point p/ pawn.) The pieces get evaluated for how effective it is, and therefor a rook in a open file is better then a rook in the Closed file because it controls more squeres. Comps. do not obey the center because it was told to, but rather because the pieces are more effective in the middel of the bord. However, the comp. are fantastic to see an evaluate a move based by how the position would look like playing the best* moves f.x. 20 half moves ahead playing the different moves.

That was a little off-topic, but the idea is good, but may be difficult to succed.