Ask HN: Is the Turing Test still the gold standard of AI?
Is it still the gold standard to measure if we "have made it"?
Is it still the gold standard to measure if we "have made it"?
A few year ago, playing chess was a task difficult enough to require intelligence, until someone solved it and it became only a brute force trick. Later, playing go was a task difficult enough to require intelligence, until someone solved it and it became only a brute force trick. You can add image classification, translation, whatever task or test you want to add. When it is unsolved it requires natural or artificial intelligence. Once it is solved is only a trick. https://en.wikipedia.org/wiki/AI_effect
We only agree that humans are intelligent, so the only test that we can make to ensure that a computer is intelligent is that the computer can fake a human. Anything else will be dismissed as a trick. And when we have a computer that can pass the Turing test it will dismissed as a trick anyway.
The problem is that making a computer that can fake a human is more complicated than we expected (or perhaps less complicated than we expected, it depends on whom you ask). So the current advancement are in small subtask.
When we have more computing power, more subtask solved, more clever ideas to solve subproblems and more time to make attempts, we will be able to make a computer that pass the Turing test. (Or not, nobody is sure, I guess we can.)
How about a nice game of chess?