Chess engines play a different kind of game. The best engines are 400 elo points better then the best human players. Yet, they are notoriously bad in openings and closed games. Both situations require the capacity to make plans, not just merely calculating power. I am not interested in evaluation of chess engines at the start of the game. To play the Sicilian is an inaccuracy in the eyes of some chess engines, right.
Secondly, chess engines have an unhuman way to find their best moves. Brute force calculation. We can not do that. They know their way in their variants, but how can we expect us to do the same? If the best human players can not see and understand their moves, how can we expect we will? Compared to the best human players you and I have the same rating. When the situation becomes too complex for us, we will start making mistakes. If we will follow the moves of the chess engines blindly, then will we end up in situations we will not be able to understand.
I have found that there is also a difference between what is practical and "highest eval." The "highest eval" move sometimes put you in a position that has illogical value for you, and the opponent's plan is so easy to find. A human player will quickly go wrong in such a position.