Lately, I've been watching replays of the AlphaStar AI's games against the top human StarCraft 2 players. The guy who does commentary on them is funny, and the matches themselves are often fascinating: https://www.youtube....ser/LaughNgamez
It's interesting to see AlphaStar make odd mistakes sometimes that reveal an apparent lack of common sense, which humans take for granted. Much more often, AlphaStar doesn't make these kinds of mistakes and plays about the same as a human. In time, the odd mistakes will stop happening, and it will be unbeatable.
This makes me realize that a similar thing will happen with respect to the Turing Test. Machines won't go from passing the Test 0% of the time one day, and 100% of the time the next day; they will creep upwards from 0% to 10% to 20% and so on. For this reason, I predict the first machine to pass the Turing Test will fail subsequent Tests, and while the initial event will be huge news, it will more than anything make us rightly question the value of the Turing Test as a yardstick of intelligence.