How computers see chess
It's interesting to try to see the game as the computer does. Broadly speaking, the computer evaluates all possible moves for a set depth and then assigns a value to each resulting position. It decides that the move giving the highest minimum value at the ends of the analysis tree is the best move. If you look at the point scores assigned by a chess program as it analyses a finished game, you will see that a move rarely increases the score for the side moving. When Black moves, whatever they do, the best possible outcome is that Black's position is maintained. After they move, Black's score can remain equal or almost the same (good move), go down by 0.1 (inaccuracy), go down by 0.2 to 1.0 (mistake), or go down by 2 or more points (blunder).
Admittedly, there are cases where the score goes up on the move, but I think these are either due to computer glitches or to the event horizon extending by one step. For instance, I have seen cases where a move was forced, yet it caused the position evaluation to go down.
Does this mean there is no such thing as a move that improves your position? From the computer's point of view, the answer is "yes". In its view, the only moves that improve your position are inferior moves made by your opponent. This is at odds with our normal perception of chess, ie that people make "strong" moves that change a position from equal to superior, or even "winning", for the side making the move. These are the moves that attract exclamation marks from human annotators. Yet in truth even such moves do not tilt the position from equal to winning. It's true that there are cases where the losing side has not made any mistakes, but they almost certainly made one or more inaccuracies, since with best play chess should be a draw. The moves annotated with exclamation marks are preceded by moves marked '?' made by the other player.
So is there such a thing as a brilliant move, or is this an illusion? Who is right, the silicon monster, or the carbon-based life-form?
I guess the answer depends on how you look at chess. If you look at chess as a mechanical puzzle to be solved, then the computer is right. If you look at it as a battle between two fallible and creative human beings, each striving to exploit the other's weaknesses, then the computer is wrong. In any case, it is impossible for the computer to evaluate whether a move is hard to see for a human or not. Essentially, a brilliant move, such as a sacrifice that appears not to achieve anything, is one that goes counter to conventional thinking.
Consider also that the computer will never lay a trap, as human players often try to do. That is, it will never lay a trap if the opponent can make a response move that improves their position even slightly. Unlike a human, the computer is incapable of seeing that the right response move can be hard to see for its opponent. As most people know, chess is not just a logical game - there is an element of psychology. That element is totally absent from the computer's purview and constitutes one of the few advantages a human has over a computer. Note, however, that this only applies to playing humans - if you are playing against an electronic brain then forget about trying to use psychology!
Note that there is a style of chess called "anti-computer chess" where the human player avoids all tactics and complications. It involves playing positional moves that aim for a long-term advantage that the computer cannot see in its game tree search. This style of play is designed to exploit the computer's weakness in positional evaluation.
Because it uses fixed rules of evaluation, the computer is unable to evaluate positions as well as a master. Unlike the master, it has zero understanding of chess. It has no plans, no notions of what to do, and does not 'think' in the ordinary sense. All it does is relentlessly calculate possible moves for as far ahead as time allows. This brute force approach is not like human thinking, but it is deadly.