In 1996 Gary Kasparov, world chess champion, played Deep Blue, IBM's machine learning computer and defeated it handily. But the 1997 rematch is a lesson in both the power and fragility of confidence in human performance.
In game one Kasparov beat the computer again, but in game two of the six game series Deep Blue moved a rook in a way that Kasparov could not understand and he had a mini breakdown under the realization that Deep Blue's computational power so vastly out paced his own. He accused the IBM team of cheating and never one another game in the six game series.
Years later one of the IBM engineers would tell reporter Nate Silver that the mysterious rook move that threw Kasparov's confidence was a glitch in the system, and completely random.