I think the main problem with the Turing Test as implemented in the reality we have started to become apparent in the 1990s. The bots that were the most convincingly human were those that were taciturn and unresponsive in a human-like way. Mastering that was not too difficult for them but nobody was all that impressed by it. Now that they've learned some of the human art of spewing bullshit — along with some other tricks — the situation is not so different in terms of intelligence but obviously it's much more marketable.