I referred earlier to the match Hitech-Lautier.  Here is the rather
beautiful 2nd game of that match.  Lautier remarked after the match
that "the computer saw everything he was trying to do".

White : Hitech                Black: J. Lautier
Cannes,France;  February 23, 1987.
        Game 2

  1  e2e4   e7e6
  2  d2d4   d7d5
  3  b1d2   c7c5
  4  g1f3   a7a6
  5  e4d5   e6d5
  6  d4c5   f8c5
  7  d2b3   c5d6
  8  f1d3   g8e7
  9  e1g1   e8g8
 10  f3d4   e7g6
 11  d1h5   b8c6
 12  d4f3   c6e7
 13  f1e1   h7h6
 14  b3d4   c8d7
 15  c1e3   d6f4
 16  e1e2   d8c7
 17  a1e1   a8e8
 18  d3g6   f7g6
 19  e3f4   c7f4
 20  h5e5   f4f7
 21  e5e7   e8e7
 22  e2e7   f7f6
 23  e7d7   g6g5
 24  e1e7   f8c8
 25  d7b7   g5g4
 26  f3e5   h6h5
 27  d4f5   g8h8
 28  f5g7   c8c2
 29  e7e8   h8h7
 30  g7f5   Resigns

Apologies for the "retarded algebraic" notation used in the score of the
game.  This was back in the early days of Hitech when we were only working
on things that made a difference.  The game was played as part of a two
game exhibition match.  It was played on the day that the GM tournament
started and GM John Fedorowicz was quoted in the press as saying that
Karpov could not have played the white pieces as well.  Of particular note
is 10. Nfd4!! which is a major improvement (on the fly) over the book Nbd4.
Yes, this is the wrong knight, but it allows the queen to participate in
a very meaningful attack on the king, which playing Nbd4 does not.
The combination started with 18. B:g6! is correct and winning.  After
20. Qe5!! White wins at least a pawn.  The attempt to avoid this by
20.--Qf7 runs into the queen sac 21. Q:e7! after which everything becomes
very easy.